Arthritis & Rheumatism (Arthritis Care & Research) Vol. 49, No. 3, June 15, 2003, pp 314 –320 DOI 10.1002/art.11124 © 2003, American College of Rheumatology ORIGINAL ARTICLE Operant Behavioral Treatment of Fibromyalgia: A Controlled Study KATI THIEME, ERIKA GROMNICA-IHLE, AND HERTA FLOR Objective. To evaluate the efficacy of operant pain treatment for fibromyalgia syndrome (FMS) in an inpatient setting. Methods. Sixty-one patients who fulfilled the American College of Rheumatology criteria for FMS were randomly assigned to the operant pain treatment group (OTG; n ⴝ 40) or a standardized medical program with an emphasis on physical therapy (PTG; n ⴝ 21). Pain assessments were performed before, immediately after, 6 months after, and 15 months after treatment. Results. The OTG patients reported a significant and stable reduction in pain intensity, interference, solicitous behavior of the spouse, medication, pain behaviors, number of doctor visits, and days at a hospital as well as an increase in sleeping time. Sixty-five percent of the OTG compared with none of the patients in the PTG showed clinically significant improvement. Conclusion. These results suggest that operant pain treatment provided in an inpatient setting is an effective treatment for FMS, whereas a purely somatically oriented program may lead to a deterioration of the pain problem. KEY WORDS. Fibromyalgia; Pain; Behavior therapy; Operant treatment. INTRODUCTION According to the 1990 classification criteria of the American College of Rheumatology (ACR), fibromyalgia syndrome (FMS) is defined as widespread pain in combination with tenderness at 11 or more of 18 specific tender point sites (1). The most frequent additional symptoms are chronic exhaustion, reduced ability to cope with stress, sleep disorders combined with a pathologically altered sleep pattern (2,3), and concentration problems. More than 40% of FMS patients also have thyroid problems and primary Sjögren’s syndrome (4). FMS is accompanied by many vegetative and functional disorders, such as irritable bowel syndrome, classic migraine, or tachycardia. Pharmacologic treatments of FMS include nonsteroidal antiinflammatory drugs (NSAIDs), muscle relaxants, tricyclic antidepressants, and serotonin antagonists, which have direct or indirect consequences on the hypothalamic–pituitary–adrenal (HPA) axis. The therapeutic goal is to Kati Thieme, PhD, Erika Gromnica-Ihle, MD, Herta Flor, PhD: University of Heidelberg, Mannheim, Germany. Address correspondence to Kati Thieme, PhD, Department of Psychosomatic and Psychotherapeutic Medicine, University of Heidelberg, Central Institute of Mental Health, J 5, D-68159 Mannheim, Germany. E-mail: email@example.com. Submitted for publication January 5, 2002; accepted in revised form June 13, 2002. 314 regulate the HPA axis and thus reduce such symptoms of FMS as pain, weakness, sleep disorders, and psychological stress reactivity (5). A metaanalysis of 9 studies on the efficacy of tricyclic antidepressant medication (6) showed that antidepressives have the most significant effect on quality of sleep. Stiffness and sensitivity to pressure had the lowest effect size. Thirty percent of the patients were responders (⬎50% improvement of symptoms). Serotonin antagonists showed a responder rate of 30 –50% (7,8), NSAIDs had a rate of 37% (9). The majority of the pharmacologic treatment studies indicate short-term effects (4 –9 weeks). There is a need for more placebo-controlled double-blind studies with longer outcome assessments (6). Nonpharmacologic interventions, such as cognitive-behavioral treatment programs, showed significant changes in pain-related convictions of control and self instructions, reductions in pain intensity and interference, as well as improvements in emotional variables (10 –13). A study on the efficacy of an interdisciplinary outpatient cognitivebehavioral treatment revealed significant improvements in pain severity, interference, life control, affective distress, depression, perceived physical disability, as well as a reduction of solicitous spouse behaviors (13). A clinically significant reduction in pain severity, measured with the reliability of change index, was found in 42% of the sample and was maintained over a period of 6 months (13). A metaanalysis (14) of 49 FMS treatment outcome studies compared the efficacy of pharmacologic and nonpharmacologic treatments (cognitive-behavioral treatment and Operant Behavioral Treatment of Fibromyalgia physical therapy) with respect to physical status, FMS symptoms, psychological status, and functional ability. Antidepressive medication showed significant improvements in physical status and subjective FMS symptoms. Cognitive-behavioral pain treatment yielded significant improvements in all 4 criteria as compared with physical therapy. Cognitive-behavioral treatment was more effective with respect to improvement of subjective FMS symptoms and daily functioning than pharmacologic treatment. This metaanalysis suggests that optimal treatment for FMS includes cognitive-behavioral methods. Although operant behavioral treatment has been widely applied to a variety of chronic pain syndromes (15,16), it has so far not been systematically tested with FMS patients. Operant treatment involves interventions such an increase of activity levels at home and at work, inclusion of significant others to reduce reinforcement of pain behaviors, and the management of pain-related medication (17). Because chronic pain is invariably accompanied by learning-related changes that lead to the development of pain behaviors, and because FMS patients have been shown to display pain behaviors and high levels of interference related to their pain, an operant treatment approach might also be effective with this group (15,16). For example, Turk et al (18) identified a dysfunctional subgroup of patients with high levels of pain and pain-related interference as well as high solicitous spouse behaviors who might especially profit from operant procedures. The goal of the present study was the evaluation of an inpatient program of operant pain treatment for patients with FMS. The operant pain treatment was compared with a standard inpatient medical treatment program that had physical therapy components. The effects were tested before treatment, immediately after treatment, and at 6- and 15-month followups. In addition, the psychosocial subgroups that were described by Turk et al (18) were determined and it was tested to what extent the dysfunctional subgroup profited especially from operant pain treatment. 315 Table 1. Demographic and clinical data of the patients* Age, years M SD Range Duration of the pain, years M SD Range Pain zones, n M SD Range Pain intensity, pain diary M SD Range Marital status, n (%) Single With significant other With significant other and children Education, n (%) ⱕ10 years 11–13 years University degree Occupational status, n (%) Working Unemployed Workers’ compensation Duration of occupational activity, years M SD Range Operant pain treatment group n ⴝ 40 Physical therapy group n ⴝ 21 46.61 8.67 16–64 48.51 7.51 17–62 17.05 7.12 0.6–33 15.57 6.32 0.6–36 6.31 2.4 4–9 6.62 1.9 5–9 4.72 0.98 3–9 4.69 0.87 3–8 6 (15.00) 14 (35.00) 20 (50.00) 2 (9.53) 13 (61.90) 6 (28.57) 18 (45.00) 12 (30.00) 10 (25.00) 11 (52.38) 5 (23.81) 5 (23.81) 11 (27.50) 20 (50.00) 9 (22.50) 6 (28.57) 10 (47.62) 5 (23.81) 26.00 7.72 0–32 25.40 8.82 0–34 * M ⫽ mean; SD ⫽ standard deviation. PATIENTS AND METHODS Patients. Sixty-one female FMS patients were recruited from the regular patients of a hospital for rheumatic disorders at Berlin-Buch. Inpatient treatments for rheumatic disorder in special hospitals for rheumatic disorders are common in Germany and the sample can be viewed as representative of the German FMS population. The patients signed informed consent and were randomly assigned to either an operant treatment group (OTG) or a standard medical treatment group with physical therapy (PTG). The 2 treatments were not provided at the same time and provision of the treatment was counterbalanced for time of year. The study was approved by the ethics committee of Humboldt University. All patients had an FMS diagnosis based on the ACR criteria (1). Exclusion criteria for the study were inflammatory cause of the pain; neurologic complications; duration of pain less than 4 months; pregnancy; another severe disease such as a tumor, liver, or renal disease; major psychiatric disorders; problems with the German language. The groups were not significantly different with respect to age, duration of pain, pain intensity, or the number of distinct pain areas as indicated in the pain assessment interview (see Table 1). About half of the patients were unemployed, 20% received workers’ compensation. Initially, 63 patients had begun the inpatient treatment. Two patients of the OTG dropped out after 4 days of treatment. The reasons for drop out were a severe depressive episode in 1 patient and a bipolar disorder diagnosis in the second patient. Thus, 61 patients completed the entire treatment. Methods. Medical somatic assessment. The somatic assessment included a standardized internal medicine and rheumatologic program, including assessment of blood values, a neurologic diagnosis, evaluation of tender points, and imaging methods such as radiographs or magnetic resonance imaging. 316 Verbal-subjective pain assessment. The Multidimensional Pain Inventory (MPI) (15,19,20) was used to determine pain intensity, interference of the pain, life control, affective distress, social support, and self efficacy. In addition, significant other responses (i.e., solicitous, punishing, and distracting behaviors) and the general activity level of the patients were assessed. The patients also completed a pain diary where pain intensity was recorded hourly on visual analog scales. Behavioral pain assessment. The intake of medication, amount of daily activities, and sleep behavior were assessed in the pain diary. At the followup, assessments were from the patients’ medical records at the rheumatology outpatient clinic the patients attended every 6 weeks. Activity levels were recorded using the respective scales of the MPI and pain behaviors were assessed with the Tübingen Pain Behavior Scale (TBS) (21), which was completed by only 41 significant others. The low number of TBS values was due to the fact that 5 patients did not have spouses and 15 of the spouses refused to complete the scale. The number of doctor visits and the number of days at a hospital in the 12 months before and after treatment was taken from the medical records. Treatment. A standardized operant pain treatment program was employed based on the manual devised by Flor and Birbaumer (15,22). It comprised time contingent intake and reduction of medication; increase of bodily activity; reduction of interference of the pain in the family, at work, in leisure time, and in social activities; reduction of pain behaviors in dealing with the medical system; and training in assertive pain incompatible behavior. The patients engaged in role play to reduce pain behaviors and to increase healthy behaviors. The entire group used a reinforcer plan that consisted of the delivery of a red card when pain behaviors were shown and a green card when healthy behaviors were displayed. The patients also trained increases in activity levels assisted by a physical therapist. The reduction of medication was instituted immediately following the assessment phase and was based on a physician-coordinated individual time contingent interval plan. The patients received homework that included increase of activities and reduction of pain behaviors when dealing with members of the inpatient clinic; this homework was extended to the family during family visits on the weekends. The operant pain program was conducted in groups of 5 to 7 patients. The treatment team consisted of a psychologist, a rheumatologist, a physical therapist, and a member of the nursing personnel. The standardized physical therapy treatment included 7 different types of mainly passive physical therapy exercises (e.g., muscle relaxation with thermotherapy, mud bath, concentrated relaxation and light movement therapy in warm water) as usually applied to this group of patients in this type of clinic setting in Germany. In addition, the patients received antidepressants. The treatment was conducted in open groups and the team consisted of the rheumatologist, a physical therapist, and the nursing personnel. The OTG and the PTG teams were composed of different personnel to prevent experimenter bias. Both treatments began with a patient education module Thieme et al (4 hours) provided by a rheumatologist in which the patients and significant others were informed about the symptoms, possible etiology, and possible somatic treatment strategies of FMS. Both treatments were employed daily over the course of 5 weeks and comprised 71 hours of treatment plus 4 hours of patient education. Statistical analysis. We employed repeated measures analyses of variance with the treatment groups as between factors and the assessment times as within factors. This was followed by Bonferroni corrected t-tests. The effect sizes for both treatment groups were computed based on the formula PTG (MeanT2– 4) – OTG (MeanT2– 4)/OTG (standard deviationT1) (23). The reliable change (RC) index (24) was computed as RC ⫽ X2 – X1/S1 公1 – Rxx, where X1 ⫽ T1 value, X2 ⫽ T4 value, S1 ⫽ standard deviation, and Rxx ⫽ test-retest reliability. The RC index is an empirical measure for the responder rate of the respective treatment. To assess predictors of outcome of the various treatments, the initial values of the responders and nonresponders were compared. To assess subgroups of FMS and to compare them within the course of treatment, a cluster analysis was performed that included both treatment groups. RESULTS Pain intensity and interference. A significant interaction for group ⫻ time was found for pain intensity (F[3,177] ⫽ 33.82, P ⬍ 0.001) as well as interference (F[3,177] ⫽ 35.84, P ⬍ 0.001). A significant improvement occurred only in the OTG (intensity: F[1,59] ⫽ 33.74, P ⬍ 0.001; interference: F[1,59] ⫽ 53.28, P ⬍ 0.001), with the OTG showing significantly reduced pain intensity (t ⫽ – 6.83, P ⬍ 0.001) and interference (t ⫽ –7.87, P ⬍ 0.001) compared with the PTG at all 3 posttreatment assessment times. By contrast, the PTG showed a significant increase in pain intensity and interference immediately posttreatment (pain intensity: t ⫽ – 8.92, P ⬍ 0.001; interference: t ⫽ –7.41, P ⬍ 0.001) as well as at the 6-month (pain intensity: t ⫽ –5.99, P ⬍ 0.001; interference: t ⫽ –5.34, P ⬍ 0.001) and 15-month followups (pain intensity: t ⫽ –5.69, P ⬍ 0.001; interference: t ⫽ – 6.98, P ⬍ 0.001). Pain intensity (1.68 –2.14) and interference (1.97–2.5) showed the highest effect sizes (see Table 2) of the operant treatment. In general, the effect sizes of the OTG increased over time. Emotional and cognitive variables. Affective distress (F[3,177] ⫽ 22.65, P ⬍ 0.001), life control (F[3,177] ⫽ 15.02, P ⬍ 0.001), and self efficacy (F[3,177] ⫽ 18.72, P ⬍ 0.001) all showed a significant interaction of group ⫻ time. Affective distress, life control, and self efficacy changed significantly over time only in the OTG (T2: F[1,59] ⫽ 28.61, P ⬍ 0.001; T3: F[1,59] ⫽ 15.49, P ⬍ 0.001; T4: F[1,59] ⫽ 14.69, P ⬍ 0.001), with the OTG displaying significantly lower affective distress (t ⫽ – 6.19, P ⬍ 0.001), higher life control (t ⫽ 7.09, P ⬍ 0.001), and higher self efficacy (t ⫽ 10.72, P ⬍ 0.001) than the PTG. By contrast, the PTG showed a significant increase of affective distress immediately after treatment as well as at the 6- and 15-month followups (t ⫽ –3.62, P ⬍ 0.002), Operant Behavioral Treatment of Fibromyalgia 317 Table 2. Means, standard deviations, and effect sizes of the dependent variables in the operant and physical therapy groups* Dependent variables MPI scales Pain intensity Interference Life control Affective distress Social support Self efficacy† Punishing responses Solicitous responses Distracting responses Household activities Activities outside the house Leisure activities Total activity scale Diary/medical record data Number of doctor visits Number of days at the hospital Sleep behavior, hours Medication Tübingen Pain Behavior Scale T1 T2 T3 T4 Groups M (SD) M (SD) M (SD) M (SD) OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG 4.43 (0.98) 4.34 (1.11) 4.35 (1.01) 4.43 (0.91) 3.11 (1.27) 3.09 (1.19) 3.69 (1.33) 3.72 (1.57) 3.82 (1.88) 3.79 (1.56) 3.05 (1.53) 2.66 (1.46) 1.37 (1.61) 1.31 (1.46) 3.30 (1.9) 3.29 (1.3) 3.34 (1.85) 3.36 (1.17) 3.98 (1.21) 4.52 (1.30) 1.12 (1.22) 1.77 (1.35) 2.50 (0.91) 2.44 (0.87) 2.53 (0.78) 2.89 (0.91) 3.82 (0.96) 5.47 (1.06) 3.29 (1.02) 5.28 (0.86) 3.73 (0.76) 2.43 (0.94) 2.54 (1.03) 4.46 (1.48) 3.49 (1.59) 3.40 (1.95) 3.67 (0.73) 1.76 (1.14) 1.27 (1.61) 1.17 (1.43) 2.78 (1.53) 3.48 (1.41) 3.41 (1.29) 3.39 (1.07) 3.93 (1.06) 4.50 (1.30) 1.18 (1.01) 1.77 (1.35) 2.81 (0.89) 2.47 (0.86) 2.65 (0.65) 2.90 (0.92) 3.66 (1.22) 4.85 (0.86) 2.96 (1.18) 4.83 (0.72) 3.85 (0.91) 2.64 (0.77) 2.38 (1.29) 4.47 (1.65) 3.48 (1.55) 3.39 (1.93) 3.75 (1.27) 1.90 (1.09) 1.12 (1.54) 1.22 (1.53) 2.75 (1.33) 3.68 (1.50) 3.29 (1.34) 3.55 (0.98) 4.05 (1.14) 4.52 (1.30) 1.28 (1.26) 1.77 (1.35) 2.74 (0.97) 2.47 (0.86) 2.68 (0.81) 2.90 (0.93) 3.18 (1.27) 5.28 (0.83) 2.79 (1.37) 5.33 (0.81) 3.95 (0.95) 2.18 (0.85) 2.46 (1.28) 4.78 (1.60) 3.33 (1.31) 3.99 (1.76) 4.07 (0.92) 1.41 (0.93) 0.86 (1.18) 1.22 (1.64) 2.67 (1.36) 3.98 (1.69) 3.48 (1.14) 3.76 (1.07) 3.92 (0.97) 4.52 (1.30) 1.25 (1.04) 1.77 (1.35) 2.76 (0.97) 2.47 (0.86) 2.63 (0.70) 2.90 (0.93) OTG PTG OTG PTG OTG PTG OTG PTG OTG PTG 31.61 (19.88) 28.60 (18.40) 13.24 (22.96) 10.35 (14.76) 5.77 (2.63) 5.81 (2.27) 1.85 (2.03) 1.71 (2.17) 13.95 (2.58) 13.74 (2.63) — — — — 6.95 (1.43) 6.43 (1.25) 0.90 (1.08) 1.24 (1.38) 12.33 (2.45) 16.85 (2.47) 7.40 (1.10) 5.57 (1.83) 0.80 (0.96) 2.04 (1.80) 11.52 (1.91) 15.35 (2.30) 14.70 (9.90) 37.8 (22.03) 2.61 (6.34) 18.65 (17.82) 7.55 (1.19) 5.28 (1.59) 0.70 (1.04) 2.52 (2.23) 9.95 (2.22) 14.23 (2.12) ES 1–2 ES 1–3 ES 1–4 1.68 1.21 2.14 1.97 1.88 2.50 1.02 0.95 1.39 1.44 1.57 1.74 0.05 0.05 0.35 1.25 1.21 1.89 0.06 0.06 0.22 0.37 0.49 0.69 0.01 0.14 0.15 0.47 0.38 0.49 0.48 0.40 0.43 0.37 0.29 0.32 0.32 0.28 0.35 1.16 0.70 0.20 0.16 0.70 0.61 0.86 0.90 1.75 1.48 1.66 * T1 ⫽ pretreatment; T2 ⫽ posttreatment; T3 ⫽ 6-month followup; T4 ⫽ 15-month followup; ES ⫽ effect size; M ⫽ mean; SD ⫽ standard deviation; MPI ⫽ Multidimensional Pain Inventory; OTG ⫽ operant therapy group; PTG ⫽ physical therapy group. Effect sizes ⱖ0.7 are marked in bold indicating a high treatment effect. † Item added to German version. a significant reduction of life control (t ⫽ 4.17, P ⬍ 0.001), and a decrease of self efficacy (t ⫽ 4.99, P ⬍ 0.001). The OTG had high effect sizes for affective distress (1.44 –1.74), life control (1.02–1.39), and self efficacy (1.25–1.89) (Table 2). Again, the effect sizes increased over time. Pain behaviors. Number of doctor visits (F[1,51] ⫽ 41.037, P ⬍ 0.001), number of days at a hospital (F[1,51] ⫽ 10.75, P ⫽ 0.002), medication intake (F[3,177] ⫽ 10.69, P ⬍ 0.001), pain behavior as assessed by the TBS (F[3,120] ⫽ 28.27, P ⬍ 0.001), and sleeping behavior (F[3,177] ⫽ 9.61, P ⬍ 0.01) showed a significant interaction of group ⫻ time. Only the OTG reduced its medication intake (F[1,59] ⫽ 16.22, P ⬍ 0.001) and pain behaviors (F[1,40] ⫽ 28.27, P ⬍ 0.001) and improved its sleep behavior (F[1,59] ⫽ 9.61, P ⬍ 0.001) over time. In addition, the OTG reduced the number of doctor visits by 53.5% (t ⫽ 5.62, P ⬍ 0.001) and the number of days at a hospital by 80.3% (t ⫽ 2.84, P ⫽ 0.008) compared with pretreatment. This amounts to a reduction of $3,933 per patient per year in hospitalization costs and a reduction of $1,840 per patient per year regarding costs of primary care. By contrast, the PTG’s number of doctor visits increased by 32.2% (P ⬍ 0.001) and the number of days at a hospital by 80.2% (P ⫽ 0.034). The PTG showed a significant increase of $1,905.50 per patient per year in hospitalization costs and $442 per patient per year regarding costs of primary care. The operant pain treatment had high effect sizes (see Table 2). The OTG showed significantly reduced pain behaviors (t ⫽ –5.68, P ⬍ 0.001), improved sleep behavior (t ⫽ 6.26, P ⬍ 0.001), and significantly lower medication use 318 Thieme et al (t ⫽ – 4.37, P ⬍ 0.001) than the PTG at all posttreatment assessment times. Their medication intake related to antidepressant drugs (F[3,177] ⫽ 11.68, P ⬍ 0.001), NSAIDs (F[3,177] ⫽ 14.67, P ⬍ 0.001), and opioids (F[3,177] ⫽ 21.22, P ⬍ 0.001) was also reduced. In contrast, the PTG showed a significant deterioration of all variables immediately after physical therapy as well as at the 6- and 15-month followup times (see Table 2). The operant treatment had high effect sizes (see Table 2) especially at the 15-month followup. The activity variables of the MPI, including overall activity, leisure time activities, household activities, and activities outside the house, did not show any significant change. Significant other behaviors. A significant interaction of group ⫻ time was found only for solicitous behaviors (F[3,177] ⫽ 7.65, P ⬍ 0.001). Only the OTG experienced a significant reduction of solicitous significant other behaviors (F[1,59] ⫽ 7.19, P ⬍ 0.001) at all treatment points (t ⬎ –3.29, P ⬍ 0.001) compared with the PTG. The effect sizes were smaller (Table 2). Clinical significance of the changes. The RC index of the MPI interference scale showed a responder rate of 65% in the OTG as compared with 0% responders in the PTG (F[1,58] ⫽ 69.08, P ⬍ 0.001). The responders showed high pain intensity (F[1,38] ⫽ 5.25, P ⫽ 0.028) and interference (F[1,38] ⫽ 21.87, P ⬍ 0.001) before treatment as compared with the nonresponders. Changes in patient subgroups in the course of treatment. The cluster analysis that was based on the MPI variables revealed 3 patient subgroups that were identical to those described by Turk et al (18) and can be characterized as dysfunctional (DYS), interpersonally distressed (IP), and adaptive copers (AC). As previously reported, the DYS patients showed the highest pain intensity with strong interference and high solicitous spouse behaviors. The AC patients had the lowest values in all variables. The IP patients displayed high pain intensity and strong interference values that were somewhat lower than in the DYS group and showed, in addition, strong affective distress and high punishment from the spouse. The pretreatment group assignment of the OTG patients was 50.25% in the DYS, 17.5% in the AC, and 30% in the IP group (Figure 1). Fifteen months after the operant treatment, 27.5% of the patients were still in the DYS and 7.5% in the IP group, whereas 65% were characterized as adaptive copers. By contrast, the PTG (see Figure 1) had initially 28.57% of the patients in the DYS or the AC group and 42.86% in the IP group. Fifteen months after treatment, the AC group was reduced by 23.81% to 4.76% and the IP group was increased by 19.05% to 62.0%. No changes were apparent in the DYS group. Although the cluster assignment of the OTG and PTG was not significantly different at pretreatment (2(2) ⫽ 3.24, not significant), the cluster distribution of the OTG and PTG groups was significantly different at all 3 posttreatment assessment times (post: 2(2) ⫽ 26.99, P ⬍ 0.001; 6 months: 2(2) ⫽ 29.22, P ⬍ 0.001; 15 months: 2(2) ⫽ 32.99, P ⬍ 0.001). Figure 1. Distribution of the psychosocial subgroups: dysfunctional (DYS), interpersonally distressed (IP), and adaptive coper (AC) before treatment (T1), immediately after treatment (T2), 6 months after treatment (T3), and 15 months after treatment (T4) in the operant treatment group (OTG) and physical therapy group (PTG). The P values refer to the change in the clusters at the posttreatment assessments compared with pretreatment: *P ⬍ 0.05; ***P ⬍ 0.001. DISCUSSION The goal of this treatment outcome study was the evaluation of the efficacy of operant pain treatment for fibromyalgia compared with traditional somatic treatment. This study showed a clear superiority of the operant intervention. The most significant changes were found with respect to pain intensity, interference, affective distress, self efficacy, spouse responses to pain, medication intake, and pain behaviors. Healthy behaviors were successively increased, pain behaviors were decreased, patients learned to be more assertive, and the significant others learned to reduce their pain-increasing solicitous behaviors. Thus, pain-related interference as well as pain intensity were significantly reduced despite concurrently reduced medication intake. A similar effect was found in the number of physician visits and inpatient treatments in the year after the intervention. The clinically most relevant result was a Operant Behavioral Treatment of Fibromyalgia high responder rate of 65% in the OTG with respect to pain-related interference as compared with no responders in the PTG. The PTG deteriorated in almost all variables, a result that has previously been reported by Flor and colleagues (10,22,25) in similar studies with chronic back pain patients. In those studies as well as in the present study, a traditional inpatient treatment that is usually prescribed for patients with FMS or chronic back pain in Germany was used as the control treatment. It must be assumed that the physical therapy and medication regimen as currently applied in German inpatient hospitals increase pain and pain behaviors, probably due to the mostly passive type of treatment that may have pain-reinforcing properties. Interestingly, the deterioration of the PTG was not limited to the pretreatment–posttreatment comparison, but persisted up to the 15-month followup. The long-lasting deterioration of the PTG was unexpected and needs to be explained. It is possible that the specific type of physical therapy employed was not beneficial because FMS patients show hypo- rather than hyperfunctionality of the sympathetic and muscular system (26 –28) and would require increased activity rather than relaxing treatments. The question arises to what extent the positive effects of the operative pain treatment might be explained by the deterioration of the PTG. A closer look at the data of the statistical analysis reveals that there were significant pretreatment to posttreatment and pretreatment to followup improvements in most variables in the OTG and a high responder rate of 65%, whereas there were no significant improvements in the PTG. Thus, there were genuine clinical improvements in the OTG. The cluster analysis that was performed in this study showed a significant change in the distribution of the patients in the empirically derived subgroups of the multiaxial pain classification system as described by Turk and colleagues (18). The patients in the dysfunctional group were especially profiting from operant pain treatment and a large number of patients in the OTG changed from the DYS and IP to the AC groups. Higher values in pain and interference also led to more reduction in pain-related interference (13). It is possible that the patients who did not profit from the operant group treatment might have more benefits from a cognitive-behavioral intervention. In the Turk et al (18) study, patients with a high level of interference, high solicitous behavior of the spouse, high affective distress, and little life control profited less from cognitive-behavioral treatment. This result is opposite to the outcome of this operant treatment study where the more affected patients improved more, suggesting that different patients might do well with OTG and cognitivebehavioral treatments. Both treatment studies suggest that it is necessary to develop and evaluate subgroup-specific treatment approaches. A direct comparison of operant and more cognitive-behaviorally oriented pain treatment and their respective criteria for indication would be desirable. This study has several limitations. First, a no treatment control could not be used due to the necessity to provide inpatient treatment. However, all of the patients had longstanding FMS and it is unlikely that they would have 319 shown spontaneous improvement during their hospitalization. In addition, an inpatient setting for the treatment of FMS seems to be the exception rather than the rule in most countries and the results of this study cannot necessarily be generalized to outpatient treatments. Although outpatient operant programs have yielded excellent results with chronic pain other than fibromyalgia (15,22,25), it must be noted that the inpatient setting permits more stringent behavioral control and is more intense than the traditional outpatient groups. Taken together, the results of this study suggest that operant behavioral treatment might be an efficacious treatment approach for FMS, although criteria for indication still need to be further specified. REFERENCES 1. Wolfe F, Smythe HA, Yunus MB, Benett RM, Bombardier C, Goldenberg DL, et al. The American College of Rheumatology 1990 criteria for the classification of fibromyalgia. Report of the Multicenter Criteria Committee. Arthritis Rheum 1990;33: 160 –72. 2. Smythe HA, Moldovsky H. Two contributions to understanding of the “fibrositis” syndrome. Bull Rheum Dis 1977;28: 928 –31. 3. White KP, Speechley M, Ostbye T. The London fibromyalgia epidemiology study: comparing the demographic and clinical characteristics in 100 random community cases of fibromyalgia versus controls. J Rheumatol 1999;26:1577– 85. 4. Dohrenbusch R, Gruterich M, Genth E. Fibromyalgia and Sjögren syndrome— clinical and methodological aspects. Z Rheumatol 1996;55:19 –27. 5. Crofford LJ. The hypothalamic-pituitary-adrenal stress-axis in the fibromyalgia-syndrome. J Musculoskel Pain 1996;4:181– 200. 6. Arnold LM, Keck PE Jr, Welge JA. Antidepressant treatment of fibromyalgia: a meta-analysis and review. Psychosomatics 2000;41:104 –13. 7. Papadopoulos IA, Georgiou PE, Katsimbri PP, Drosos AA. Treatment of fibromyalgia with tropisetron, a 5HT3 serotonin antagonist: a pilot study. Clin Rheumatol 2000;19:6 – 8. 8. Smith AJ. The analgesic effects of selective serotonin reuptake inhibitors. J Psychopharmacol 1998;12:407–13. 9. Wolfe F, Zhao S, Lane N. Preference for nonsteroidal antiinflammatory drugs over acetaminophen by rheumatic disease patients: a survey of 1,799 patients with osteoarthritis, rheumatoid arthritis and fibromyalgia. Arthritis Rheum 2000;43: 378 – 85. 10. Flor H, Birbaumer N. Comparison of the efficacy of electromyographic biofeedback, cognitive-behavioral therapy, and conservative medical interventions in the treatment of chronic musculoskeletal pain. J Consult Clin Psychol 1993; 61:653– 8. 11. Nicassio PM, Radojevic V, Weisman MH, Schuman C, Kim J, Schoenfeld-Smith K, et al. A comparison of behavioral and educational interventions for fibromyalgia. J Rheumatol 1997; 24:2000 –7. 12. Nielson WR, Walker C, McCain GA. Cognitive behavioral treatment of fibromyalgia syndrome: preliminary findings. J Rheumatol 1992;19:98 –103. 13. Turk DC, Okifuji A, Sinclair JD, Starz, TW. Interdisciplinary treatment for fibromyalgia syndrome: clinical and statistical significance. Arthritis Care Res 1998;11:186 –95. 14. Rossy LA, Buckelew SP, Dorr N, Hagglund KJ, Thayer JF, McIntosh MJet al. A meta-analysis of fibromyalgia treatment interventions. Ann Behav Med 1999;21:180 –91. 15. Flor H. Psychobiologie des Schmerzes [Psychobiology of pain]. Bern: Verlag Hans Huber; 1991. 16. Vlaeyen JWS, Haazen IWCJ, Kole-Snijders AMJ, van Eek, H, Schuerman JA. Behavioral rehabilitation of chronic low back 320 17. 18. 19. 20. 21. 22. Thieme et al pain: comparison of an operant treatment, an operant-cognitive treatment and an operant-respondent treatment. Br J Clin Psychol 1995;34:95–118. Fordyce WE. Behavioral methods in chronic pain and illness. St. Louis: Mosby; 1976. Turk DC, Okifuji A, Sinclair JD, Starz TW. Pain, disability and physical functioning in subgroups of patients with fibromyalgia. J Rheumatol 1996;23:1255– 62. Flor H, Rudy TE, Birbaumer N, Streit B, Schugens MM. Zur Anwendbarkeit des West Haven-Yale Multidimensional Pain Inventory im deutschen Sprachraum: Daten zur Reliabilität und Validität des MPI-D. [The applicability of the West Haven-Yale Multidimensional Pain Inventory in German-speaking countries: data on the reliability and validity of the MPID]. Der Schmerz 1990;4:82–7. Kerns RD, Turk DC, Rudy TE. The West Haven-Yale Multidimensional Pain Inventory (WHYMPI). Pain 1985;23:345–56. Flor H, Heimerdinger K. Erfassung von Schmerzverhalten. [Assessment of pain behavior]. In: Geissner E, Jungnitsch G, editors. Psychologie des Schmerzes [Psychology of pain]. Weinheim: Psychologie Verlags Union; 1992. p. 99 –106. Flor H, Birbaumer N. Psychobiologie und interdisziplinäre 23. 24. 25. 26. 27. 28. Therapie chronischer Wirbelsäulensyndrome [Psychobiology and interdisciplinary treatment of chronic back pain]. München: GSF Forschungszentrum; 1994. Turner JA, Jensen MP. Efficacy of cognitive therapy for chronic low back pain. Pain 1993;52:169 –77. Jacobson NS, Follette WC, Revenstorf D. Psychotherapy outcome research: methods for reporting variability and evaluating clinical significance. Behav Ther 1984;15:336 –52. Flor H, Fydrich Th, Turk D. Efficacy of multidisciplinary pain treatment centers: a meta-analytic review. Pain 1992;49:221– 30. Elam M, Johansson G, Wallin BG. Do patients with primary fibromyalgia have an altered muscle sympathetic nerve activity? Pain 1992;48:371–5. Qiao Z, Vaeroy H, Morkrid L. Elektrodermal and microcirculatory activity in patients with fibromyalgia during baseline, acoustic stimulation and cold pressor tests. J Rheumatol 1991; 18:1383–9.28. Vaeroy H, Qiao ZG, Morkrid L, Forre O. Altered sympathetic nervous system response in patients with fibromyalgia (fibrositis syndrome). J Rheumatol 1989;16:1460 –5.