SEARCH

SEARCH BY CITATION

Keywords:

  • Fibromyalgia syndrome;
  • Pain behavior;
  • Predictors;
  • Operant conditioning;
  • Subgroups

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Objective

To evaluate the contributions of physical, pain-related, cognitive, stress-related, affective, and spouse-related variables to differences in pain behaviors in subgroups of patients with fibromyalgia syndrome (FMS).

Methods

One hundred forty FMS patients underwent medical, physical, and psychological evaluation. Patients and 30 pain-free controls performed a routine physical activity (window-washing task) to elicit pain behaviors with or without the presence of their spouses. The behaviors and spouses' responses during this task were videotaped and subsequently rated. Patients were classified as dysfunctional (DYS), interpersonally distressed (ID), or adaptive copers (AC) based on responses to the Multidimensional Pain Inventory. Hierarchical regression analyses were used to identify predictors of pain behaviors for the total group and subgroups of patients.

Results

Patients classified as DYS demonstrated the highest number of pain behaviors compared with those classified as ID or AC. This difference was observable when the spouse was present. Spouse responses and physical variables were significantly related to pain behaviors in the DYS and ID groups with the model accounting for 77.1% and 41.9% of the variance, respectively. In contrast, for the AC group, stress factors were the most significant predictor of pain behaviors, accounting for 22.8% of the variance.

Conclusion

The results indicate that different variables account for the presence of pain behaviors in different subgroups of patients. The data provide support for the heterogeneity of the diagnosis of FMS and have implications for treatment of subgroups of patients.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Fibromyalgia syndrome (FMS) has been defined by the American College of Rheumatology (ACR) as consisting of widespread pain of at least 3 months' duration in combination with tenderness at 11 or more of 18 specific tender point sites (1). In addition, patients diagnosed with FMS report a range of physical, cognitive, affective, stress-related, and behavioral symptoms.

Maladaptive learning processes have been viewed as essential for the development of pain behaviors and disability in many chronic pain syndromes (2–4). Pain behaviors are overt expressions of pain, distress, and suffering, such as slowed movement, bracing, limping, and grimacing (5, 6). Pain behaviors have a communicative function (5–7) and signal the presence of pain to others. A central feature of pain behaviors is that they are observable, and therefore capable of eliciting a response from significant others. From an operant conditioning perspective, increased pain behaviors result from reinforcing responses by significant others. For example, solicitous responses by significant others have been found to be positively associated with higher ratings of pain severity, more pain behaviors, greater disability, and decreased activity levels (8–10). However, pain behaviors may also be related to factors other than operant conditioning variables, for example cognitive, affective, and physical variables.

Several studies have specifically examined the presence of pain behaviors and their predictors in FMS patients (8, 11, 12). Romano et al (11) showed that spouses' attentive responses to nonverbal pain behaviors were significant predictors of physical disability in the more depressed patients, and were significant predictors of the rate of nonverbal pain behavior in patients who reported greater pain. Turk and Okifuji (12) observed that the combination of physical, affective, and cognitive predictors explained 53% of the variance of pain behaviors in a sample of FMS patients. Contrary to the results of other studies (9–11, 13), however, operant variables did not predict pain behaviors in this sample.

Several investigators have implicated the role of the hypothalamic-pituitary-adrenal (HPA) axis in FMS (14–16) as one physical factor in the chronicity of FMS. FMS patients show a dysregulation of the HPA axis with reduced growth hormone and hypocortisolism (17–19) that may influence pain thresholds (20). Low levels of cortisol may lead to heightened sensitivity to pain and other symptoms. This increased attentional focus might facilitate the opportunity for operant conditioning to occur.

To date studies on pain behaviors have focused on FMS as if the syndrome consisted of a homogeneous group. There have been suggestions that FMS patients may differ on important variables, such as responses to medication treatment, differences in biologic variables, association of depression with specific cytokine abnormalities, the presence of antipolymer antibody (21), and psychosocial status (22). The failure to consider subgroup differences may contribute to the inconsistency in predictors of pain behaviors, a lack of understanding of FMS, and ultimately inadequate and inappropriate treatment.

Turk and his colleagues (22) used cluster analytic procedures to identify subgroups in chronic pain patient populations based on their responses to the Multidimensional Pain Inventory (MPI) (23). One group, labeled dysfunctional (DYS), exhibited the highest level of pain, emotional distress, and disability. A second group, labeled interpersonally distressed (ID), reported significantly lower levels of pain, disability, and marital satisfaction than the other 2 subgroups. The significant others of ID patients showed a higher level of negative responses to the patients' expressions of pain. The third group, adaptive copers (AC), showed low pain intensity, emotional distress, and interference of pain with daily life and activities.

The primary objective of the present study was to confirm the role of significant others in the presence of pain behaviors for FMS patients. In addition, the role of physical and psychosocial variables in predicting pain behaviors was evaluated. Finally, differences in the prevalence of pain behaviors among subgroups of patients were of interest. Specifically, we examined the following hypotheses: 1) FMS patients will demonstrate a greater number of pain behaviors than pain-free controls when they perform a task requiring physical activity; 2) the pain behaviors of FMS patients will increase during a physical task when their spouse is present; 3) FMS patients classified as DYS will demonstrate a larger number of pain behaviors than those classified as AC or ID; 4) operant (spouse-response) variables will be the best predictors of pain behaviors in the DYS but not in the AC and ID groups; and 5) lower cortisol levels may contribute to enhanced pain sensitivity and thereby more pain behaviors.

SUBJECTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Subjects.

One hundred forty FMS patients and 30 pain-free controls and their spouses participated in the study. The patients were recruited from consecutive, married FMS patients who attended internal medicine and rheumatology outpatient clinics in eastern Germany. All patients received an intensive physical examination and they all fulfilled the ACR criteria for FMS (1). In addition to the diagnosis of FMS, the following inclusion criteria were used: 1) pain for a period of at least 6 months; 2) married; 3) willingness of the spouse to be involved in the assessment (only 2 persons were excluded due to lack of spouse willingness); 4) adequate cognitive ability; and 5) fluency in the German language. The average age of the patient sample was 46.71 years. More than 50% of the patients were unemployed or on worker's compensation (see Table 1). No male FMS patients (n = 5) were willing to participate. The mean duration of pain was 10.53 years, and on average 6.66 body regions were indicated as being painful by the patients versus 0 in the control group. The average number of positive tender points was 15.01 and the mean pain intensity on a scale ranging from 0 = “not at all painful” to 10 = “extremely painful” based on the Manual Tender Point Survey (MTPS) (24) was 5.97; in the control group it was 0.00. The average age of the spouses was 46.71 years.

Table 1. Demographic and clinical data of the patients (n = 140), significant others (n = 140), and healthy controls (n = 30)
 PatientsSpouses of patientsHealthy controlsSpouses of healthy controls
  • *

    n = 46.

Age, years    
 Mean46.7149.2848.6255.01
 SD10.5811.4212.8810.23
 Range21–6621–6821–6721–68
Duration of pain, years    
 Mean10.535.21* 7.53
 SD9.744.83 5.08
 Range0.6–332.0–10.0 2.0–12.0
Pain regions (n = 10)    
 Mean6.663.0* 2.0
 SD2.562.4 1.2
 Range1–100–6 0–4
Number of tender points (maximum n = 18)    
 Mean15.013.0*0.02.0
 SD1.261.90.00.6
 Range11–180–6 0–4
Pain intensity of tender points (range 0–10)    
 Mean5.972.9*0.02.5
 SD2.281.30.01.1
 Range1.00–9.520.0–4.3 0.0–3.8
Duration of occupational activity, years    
 Mean24.6525.3625.8828.06
 SD11.5810.2110.4312.43
 Range0–450–450–450–45
Occupational status, n (%)    
 Working46 (32.8)94 (67.1)12 (40.0)10 (33.3)
 Unemployed60 (42.9)26 (18.6)12 (40.0)12 (40.0)
 Workers' compensation24 (17.1)10 (7.1)3 (10.0)3 (10.0)
 Retired6 (4.3)6 (4.3)1 (3.3)3 (10.0)
 Student4 (2.9)4 (2.9)2 (6.7)2 (6.7)

To match patients to a comparable group, the pain-free controls were recruited from the circle of pain-free friends of patients who agreed to participate. The average age of the controls was 48.62 years and they were comparable in sex and social status (see Table 1). Fifty percent of the controls were working, 10% were on disability (mostly due to cardiovascular problems), and 40% were unemployed (this is consistent with the unemployment rates in eastern Germany). None of the controls reported the presence of any pain at the time of participation in the study.

Procedure.

This 1-year study adhered to the guidelines of the Declaration of Helsinki, and the local Institutional Review Board approved the study. The study was described to all participants (patients, their spouses, and controls) and all signed informed consent. The instructions noted that a cannula would be inserted (patients and controls) to permit sampling of blood, and that movement-related variables and the ability to perform activities would be tested by using a household activity. A cannula was inserted by an anesthesist 20 minutes before blood sampling and was removed before the window washing. Endocrine parameters were assessed immediately prior to the observation period. The Institute for Laboratory Medicine and Pathochemistry (Charité, Humboldt University, Berlin) performed the assays of adrenocorticotropic hormone (ACTH), cortisol, growth hormone (STH), and thyroid hormones (T3 and T4). The assays were all performed at the same time to reduce day-to-day variability in the assay. The technician was blinded to the sample source and intervention code. The assays for the endocrine mediators were highly specific for the specific entity being tested. The assayed hormones had no intrinsic seasonal variability in the human system.

Each participant (patient and control) was asked to wash a window (3 × 4–foot rectangle, angled so as to require squatting during part of the washing process; see Figure 1) for a period of 8 minutes. This task was selected because it required vigorous physical activity and was representative of common household activities with which the participants were familiar. The participants performed the task for 4 minutes in the presence and 4 minutes in the absence of their spouses. The presence or absence of the spouse during the performance period was not explained. In half of the participants, the spouse was present during the first 4 minutes, in the other half in the second 4 minutes of the trial. The entire sequence was videotaped and permitted observation not only of the participants' behaviors but their spouses' responses as well. Presence and pain intensity of tender points were assessed before and after the behavioral task. During the period in which the patient and the controls performed the window washing in the spouse-absent phase, the spouse completed pain-related questionnaires in an adjoining room.

thumbnail image

Figure 1. Photograph depicting the window washing task.

Download figure to PowerPoint

Somatic variables.

The number of somatic symptoms was derived from patients' medical records. The Fibromyalgia Impact Questionnaire (FIQ) (25, 26) was used to assess symptoms of FMS, including stiffness, fatigue, morning tiredness, and physical impairment.

The MTPS provides a standardized protocol for assessing the presence of positive tender points and the severity of pain at each tender point (24). The protocol follows a standard order and force of palpation and specifies the position of both the patient and physician throughout the exam. Patients were asked to rate (on a scale of 0–10) the severity of pain they experience after each of the standard tender point locations was palpated with 4 kg of force. The total number of positive tender points was recorded and the mean pain intensity of the tender points was derived by summing the patients' responses and dividing the total by 18.

Psychometric assessment.

All standardized questionnaires have been used widely in chronic pain studies (13, 22, 27–29) because of their excellent psychometric properties. The State-Trait Anxiety Inventory (STAI) (30) consists of 20 items used to assess dispositional and situational anxiety. Current symptoms of depressed mood were assessed using the Center for Epidemiological Studies Depression Scale (CES-D) (31, 32).

The FIQ (25, 26) is a self-report questionnaire that contains questions regarding the number of days felt good, missed work days, job difficulty, pain, anxiety, and depression along with a 10-item subscale of physical functioning. The West Haven-Yale Multidimensional Pain Inventory (MPI) (23, 33) is a 60-item questionnaire that asks patients to rate their pain intensity, interference of the pain, life control, affective distress, social support, self-efficacy, significant other responses (i.e., solicitous, negative, and distracting behaviors), and general activity level.

Cognitive variables were assessed using the Pain-Related Self-Statements Scale (PRSS) (34) with the subscales active coping (e.g., I can handle my pain) and catastrophizing (e.g., I am a hopeless case). The Brief Stress Scale (BSS) (34) was used to assess everyday stress. This measure has 4 subscales: marital distress, stress in everyday life, social stress, and work-related stress. In addition, a total stress score was computed.

Behavioral observation.

The prevalence and frequency of the occurrence of each pain behavior was coded from the videotape in 10-second epochs during the entire 8-minute trial using the Tübingen Pain Behavior Scale (TBS) (34). The TBS rates the presence of 11 pain behaviors (e.g., groaning, slowed movements, pain-based refusal of activities) on a 0–2 scale (0 = none, 1 = sometimes, and 2 = always). The coding was carried out separately for the conditions “presence of spouse” and “absence of spouse,” leading to an assessment of 24 sequences of 10 seconds each for every condition. The total number of pain behaviors was calculated by summing the absolute frequencies of the individual pain behaviors observed during the task. Summed scores for the conditions in which the spouse was present or absent were computed. The videotaped spouse responses were rated as negative, solicitous, or distracting based on the MPI (23, 33), using 8 30-second observation sequences. The raters assessed the behaviors based on specific examples provided to them such as “spouse helped” and “spouse told patient to stop” for solicitous; “spouse sat quietly” and “spouse attended to other things” as punishing/ignoring; and “spouse tried to cheer patient up” and “spouse talked about something else” as distracting behaviors.

For each observed spouse response, the total value of incidence per sequence was calculated. Two independent raters coded the pain behaviors and the spouse responses. The raters were blind to whether the person being observed was a patient or control. Kappa coefficients of 0.82 (P < 0.001) and 0.81 (P < 0.001) were obtained for interrater reliability of the pain behaviors and for the spouse response ratings, respectively.

Data analyses.

A cluster analysis based on MPI responses was computed to classify patients into psychosocial subgroups. To assess differences in pain behaviors between the patient and the control group and among the psychosocial subgroups, one-way analyses of variance were performed. In the case of a significant result, post-hoc comparisons were performed with Bonferroni corrections (P < 0.001).

To determine predictors of pain behaviors, multiple hierarchical linear regression analyses were computed. The following categories of independent variables were entered into the equations: demographic, pain intensity, somatic, cognitive, stress-related, affective, and spouse-related responses. Pearson's correlations were used to test associations between these variables and pain behaviors to identify the relevant parameters for the regression analysis and to reduce the number of variables. Variables were entered in the following order as sets: demographic variables (age, duration of pain, duration of occupational activity, and occupational status), somatic variables (ACTH, STH, cortisol, T3, T4), somatic symptoms (fatigue, morning tiredness, and stiffness from the FIQ), pain-related variables (pain severity before the performance trials, the pain intensity scale of the MPI, and the pain intensity scale of the FIQ), cognitive variables (catastrophizing, active coping [PRSS], and life control [MPI]), stress-related variables (marital distress, stress in everyday life, social stress, and work-related stress [BSS]), affective variables (STAI, CES-D, and the affective distress scale [MPI]), and spouse-related variables (solicitous, distracting, and negative responses derived from the observation and the MPI). Finally, the variables with a significant correlation to pain behaviors were entered in a multiple hierarchical regression to determine their contribution to the prediction of pain behaviors. This procedure was carried through for the entire sample as well as for the DYS, ID, and AC groups separately.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

Pain behaviors.

The FMS patients and controls were significantly different with respect to the frequency of pain behaviors during the window-washing task (F[1, 168] = 27.35; P < 0.001). The FMS patients showed significantly more pain behaviors than the controls both in the presence (F[1, 168] = 16.84; P < 0.001) and the absence of spouses (F[1, 168] = 30.60; P < 0.001). There were no statistically significant differences between pain behaviors when the spouse was present or not (t[139] = 0.17).

Pain behaviors in the psychosocial subgroups.

The cluster analyses replicated the 3 primary MPI profiles of psychosocial subgroups (2). The DYS group consisted of 27.86% (n = 39), the ID cluster consisted of 33.57% (n = 47), and the AC cluster consisted of 38.57% (n = 54) of the sample. There were no significant differences between the subgroups in age, but the subgroups were significantly different in pain duration (F[2, 75] = 3.24; P = 0.03). The AC group showed a significantly longer duration of pain compared with the ID and DYS groups. The latter 2 subgroups did not differ from each other (Table 2).

Table 2. Pain characteristics of the patients, broken down by psychosocial subgroup*
 DYS (n = 39)ID (n = 47)AC (n = 54)
  • *

    DYS = dysfunctional group; ID = interpersonally distressed group; AC = adaptive copers group; MPI = Multidimensional Pain Inventory.

Pain before window washing   
 Mean6.664.835.15
 SD2.522.272.29
 Range2–102–101–10
Pain (MPI)   
 Mean4.953.723.76
 SD0.761.010.86
 Range3.53–6.001.00–5.331.67–5.67
Pain duration, years   
 Mean6.425.3811.04
 SD4.193.7211.44
 Range0.5–201–162–43

The 3 subgroups were significantly different in the total number of pain behaviors (F[2, 68] = 8.43; P = 0.001) (see Figure 2). Post-hoc tests indicated that the DYS group showed significantly more pain behaviors than the ID (t[47] = −4.60; P < 0.001) and the AC groups (t[50] = −4.77; P < 0.001). No significant differences between the ID and the AC groups were identified.

thumbnail image

Figure 2. Pain behaviors in the psychosocial subgroups and healthy controls (HC) in the absence and in presence of spouse, as well as both combined as global score. DYS = dysfunctional; ID = interpersonally distressed; AC = adaptive copers.

Download figure to PowerPoint

Pain behaviors in the presence of the spouse (Figure 2) were significantly different between the subgroups (F[2, 68] = 10.33; P < 0.001). Post-hoc tests revealed significantly more pain behaviors in the DYS compared to the ID (t[47] = −3.50; P < 0.001) and AC (t[50] = −3.77; P < 0.001) groups. No significant differences were found between the ID and the AC groups. Pain behaviors in the absence of the spouse were not significantly different between the subgroups (F[2, 68] = 0.06; P = 0.94).

Predictors of pain behaviors.

Pain behaviors in the presence and absence of the spouse were summed and used as the dependent variable in the predictions. Pain behaviors of the entire sample (see Table 3) were significantly positively correlated with solicitous behavior (r = 0.31, P = 0.010), cortisol levels (r = 0.32, P = 0.03), catastrophizing (r = 0.34, P = 0.005), and stress at work (r = 0.31, P = 0.01) and significantly negatively correlated with distracting spouse behavior (r = −0.44, P = 0.001), pain intensity before the investigation (r = −0.41, P < 0.001), and coping (r = −0.28, P = 0.02). The variance in pain behaviors was accounted for by high solicitous and low distracting spouse behavior (19.2%; F[1, 139] = 3.79; P = 0.03), reduced cortisol production (17.2%; F[1, 139] = 5.22; P = 0.03), and increased stress at work (15.1%; F[1, 139] = 3.34; P = 0.04) (Table 4).

Table 3. Pearson correlation coefficients between pain, physical, cognitive, stress, and operant variables with total pain behaviors of the entire sample and of the psychosocial subgroups*
VariablerP
  • *

    DYS = dysfunctional; STH = growth hormone; FIQ = Fibromyalgia Impact Questionnaire; ID = interpersonally distressed; ACTH = adrenocorticotropic hormone; AC = adaptive copers.

Entire sample (n = 140)  
 Operant  
  Distracting behavior−0.4420.001
  Solicitous behavior0.3070.010
 Pain  
  Pain before−0.410< 0.001
 Physical  
  Cortisol0.3180.030
 Cognitive  
  Coping−0.2810.020
  Catastrophizing0.3370.005
 Stress  
  Stress at work0.3100.010
DYS group (n = 39)  
 Operant  
  Distracting behavior−0.6360.006
  Solicitous behavior0.6090.006
 Physical  
  Stiffness0.5540.014
  STH−0.6990.017
  Cortisol−0.7520.008
 Pain  
  Pain before0.5350.018
  Pain (FIQ)0.4360.046
ID group (n = 47)  
 Physical  
  ACTH0.6470.005
AC group (n = 54)  
 Stress  
  Stress at work0.4780.014
Table 4. Predictors of total pain behaviors: multiple hierarchical regression analyses of the entire sample and of the psychosocial subgroups*
VariableRR2dfFP
  • *

    df = degrees of freedom; MPI = Multidimensional Pain Inventory; DYS = dysfunctional; STH = growth hormone; ID = interpersonally distressed; ACTH = adrenocorticotropic hormone; AC = adaptive copers.

Entire sample (n = 140)     
 Spouse response     
  Distracting responses and solicitous responses (MPI)0.4380.1921, 1393.7910.033
 Physical     
  Cortisol (blood serum)−0.604−0.3642, 1385.2180.030
 Stress     
  Stress at work0.7180.5153, 1373.3370.042
DYS group (n = 39)     
 Spouse response     
  Distracting responses and solicitous responses (MPI)0.6720.4521, 387.4130.024
 Physical     
  STH (blood serum) and cortisol (blood serum)−0.878−0.7712, 377.8520.012
ID group (n = 47)     
 Physical     
  ACTH (serum)0.6470.4191, 4610.8190.005
AC group (n = 54)     
 Stress     
  Stress at work0.4780.2281, 537.0930.014

The combination of spouse-related, physical, and stress variables explained 51.5% of the variance (see Table 4). There was no predictive value shown for the demographic, pain-related, cognitive, or affective factors for pain behaviors in the entire sample.

Pain behaviors of the DYS subgroup (see Table 3) were significantly positively correlated with solicitous spouse behavior (r = 0.64, P = 0.006) and pain intensity before the performance trial (r = 0.54, P = 0.02) and were significantly negatively correlated with growth hormone levels (r = −0.70, P = 0.02), cortisol production (r = −0.75, P = 0.008), and distracting spouse behavior (r = −0.63, P = 0.006). High solicitous and low distracting spouse behavior explained 45.2% (F[1, 38] = 7.41; P = 0.02) and a reduced production of growth hormone and cortisol an additional 31.9% (F[1, 38] = 4.88; P = 0.05) of the variance in pain behaviors of DYS patients. The combination of operant and physical variables explained 77.1% of the total variance (Table 4). Demographic, cognitive, affective, and stress factors had no predictive value for pain behaviors of the DYS group.

Pain behaviors of the ID subgroup were significantly positively correlated only with enhanced production of ACTH (r = 0.65, P = 0.005) (see Table 3), which explained 41.9% (F[1, 46] = 10.82; P = 0.005) of the variance in pain behaviors (see Table 4). There was no predictive value shown by demographic, pain-related, spouse-related, cognitive, affective, or stress factors for pain behaviors of the ID group.

Pain behaviors of the AC subgroup were significantly positively correlated with stress at work (r = 0.48, P = 0.014) (see Table 3), which explained 22.8% of the variance (F[1, 53] = 7.09, P = 0.01) (see Table 4). Demographic, physical, pain-related, spouse-related, cognitive, and affective factors had no predictive value for pain behaviors of the AC group.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES

In support of the first hypothesis, the FMS group showed significantly more pain behaviors than the pain-free control group. Among the FMS patients, the DYS group had the greatest number of pain behaviors compared with the ID and AC groups. The results of these analyses partially confirm the third hypothesis of higher pain behaviors in the DYS group. This difference was especially observable in the presence of the spouse but not if the spouse was absent, confirming the second hypothesis. In accordance with the operant pain model (4) and studies on FMS and chronic back pain (7, 8, 13), pain behaviors were high in the FMS group as a whole and particularly for the DYS group. The presence of a spouse appears to serve as a discriminative cue for these behaviors.

The physical, cognitive, affective, stress, spouse-related, and pain-related variables had different predictive values for the pain behaviors of the entire sample and for the psychosocial subgroups. The prevalence of pain behaviors for the entire sample was best predicted by high solicitous and low distracting spouse behavior, a reduced cortisol production, and increased stress at work. The results regarding the role of physical variables as significant predictors of pain behaviors are in accordance with previous reports on other pain groups (35–38). Turk and Okifuji (12), for example, explained 16% of the variance of pain behaviors in a sample of FMS patients by physical factors comparable with our results accounting for 17% of the variance. Thus, physical variables are one important determinant of pain behaviors in FMS but not the sole contributor.

The relationship of reduced cortisol production on pain behaviors has not been reported previously. The role of cortisol production in relation to FMS symptoms is controversial. Bennett et al (14) suggested that reduced production of growth hormones leads to intermittent phases of heightened cortisol levels. In contrast, Griep et al (17) showed a mild hypocortisolemia in FMS patients after a dexamethasone suppression test. Heim et al (18) also demonstrated hypocortisolism in patients with FMS, rheumatoid arthritis, and bronchial asthma with comorbid posttraumatic stress disorder (PTSD) compared with nonpain patients suffering from PTSD who showed hypercortisolism (19). Hormone production is known to influence central nervous system activity. For example, release of cortisol increases sensory thresholds in healthy persons (27). The reduced cortisol production in FMS patients might contribute to their higher pain levels and pain behaviors. Higher pain sensitivity might lead to the expression of more pain behaviors. However, the relationship of cortisol and FMS symptoms, especially pain behaviors, needs further investigation.

Pain behaviors for the entire sample, and particularly for the DYS subgroup, were also predicted by operant variables. The predictor accounting for the largest amount of variance (19%) in pain behaviors for the entire sample was both solicitous spousal behavior and reduced distracting spouse responses. This dichotomy is consistent with studies (8, 39) that found solicitous behaviors as predictors for pain behaviors in the sample. The level of distracting behaviors was not the aim of earlier studies. Continuatively, it may be that solicitous spouse behavior combined with distracting behavior characterizes social support. Further behavioral observation studies that vary the spousal behavior are necessary to differ between solicitous spouse behavior reinforcing pain behavior and social support.

In the DYS group, a much higher percentage of pain behaviors was predicted than in the other groups. For this group, more solicitous spouse responses, less growth hormone and cortisol production, and more pain intensity together explained a large percentage (77%) of the variance, confirming the fourth hypothesis. The predictor accounting for the greatest amount of variance (45%) was solicitous spouse responses. This result is consistent with other studies that have shown that pain behaviors and solicitous responses by significant others are positively associated (3, 7, 9). These results replicate previous findings and further underline the relationship between pain behaviors and reduced cortisol production as a physical variable, which influences pain perception. Reduced cortisol production could lead to higher pain perception and pain behaviors, which could, in turn, enhance pain perception and response to pain.

For the ID group, heightened ACTH production was the best single predictor of pain behaviors, accounting for 42% of variance. It has been proposed that FMS should be considered a stress-related syndrome because symptoms often have their onset triggered by stress, whether psychological, infectious, or traumatic (40). Stress exerts its effects by as yet unknown central pathways that stimulate the hypothalamus to release stress hormones, with ACTH, corticotropin-releasing hormone (CRH), and antidiuretic hormone being the most important. Elevated CRH seems to alter the set point of other hormonal axes, such as the hypothalamic-pituitary-adrenal axis associated with enhanced ACTH and reduced cortisol production (7), which might be one contributing factor to pain behaviors. Born et al (41, 42) and Floretal (27) showed that higher ACTH production is related to enhanced processing of sensory stimuli, including pain (42).

The pain behaviors of the AC group were best predicted by stress at work, explaining 23% of the variance. In contrast to the other groups, stress appears to be especially important as a contributing factor to pain behaviors in the AC group. Overall, the AC showed a lower number of pain behaviors than the other 2 groups.

The different predictors of pain behaviors for the FMS subgroups support the observation that FMS is not a homogeneous syndrome (43). Different factors appear to be important in influencing how patients respond to their symptoms. The results suggest that different treatments for FMS subgroups might be useful. The significance of solicitous responses by significant others as a predictor of pain behaviors in DYS patients suggests that an operant behavioral pain treatment might be appropriate (28, 39). This treatment seems to be less warranted for the ID patients and the AC group because of their lower pain behaviors. The ID and AC patients show high pain behaviors in relation to heightened ACTH production or enhanced stress. Because the ID and AC patients show problems in coping with stress (40, 44), a cognitive-behavioral pain treatment plan would be helpful.

Although the overall sample size for the patient group appears reasonable, subdividing the total sample into 3 subgroups produced relatively small samples. Thus, the interpretation of the results of the regression analyses on the subgroups must be considered with caution. Future research is needed to replicate the predictors of pain behaviors observed in the different subgroups using larger patient samples. Furthermore, it would be necessary to compare the pain behaviors in FMS with other chronic pain conditions (e.g., osteoarthritis, back pain).

Although the FMS patients and the healthy controls showed significant differences in pain behaviors both in the presence and the absence of the spouse, and the psychosocial subgroups also showed significant differences of pain behaviors in the presence of the spouse, significant differences for pain behaviors in the absence of the spouse were not found. The absence of the differences in pain behaviors of the different subgroups under the absence of the spouse condition underlines the importance of the spouse as a discriminative cue for pain behaviors and suggests that pain behaviors are closely related to immediate reinforcement.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. SUBJECTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES