Background
Chronic widespread pain, stiffness, fatigue, sleep disturbance, cognitive difficulties, and mood disturbance are considered core symptoms of fibromyalgia. In 2008, the Outcome Measures in Rheumatology fibromyalgia working group recommended evaluation of these core symptoms in addition to measures of function and biomarkers, when available, in all clinical research into fibromyalgia [
1]. In a previous study [
2], we identified four unique symptom clusters using self-report questionnaires in a sample of 581 women with fibromyalgia using these core Outcome Measures in Rheumatology symptom domains. These clusters included: 1) a low symptom intensity group; 2) a moderate symptom intensity group with low anxiety and depression; 3) a moderate symptom intensity group with higher anxiety and depression; and 4) a high symptom intensity group.
Our results are consistent with previous studies that have also reported the presence of symptom subgroups, or
clusters, in samples of patients with fibromyalgia [
3‐
7]. For example, Yim and colleagues [
7] identified four clusters based on physical and psychological assessments and measures of pain: 1) high pain and physical and mental impairment, with low social support; 2) moderate pain and physical impairment, with mild mental impairment and moderate social support; 3) moderate pain and mental impairment, with low physical impairment and social support; and 4) low pain, with normal physical and mental function and high social support. Similarly, Docampo and colleagues [
6] identified three clusters on the basis of fibromyalgia symptoms and comorbid conditions: 1) low symptoms and comorbid conditions; 2) high symptoms and comorbid conditions; and 3) high symptoms but low comorbid conditions. These and previous studies suggest the presence of distinct symptom subgroups within samples of patients with fibromyalgia. There is usually one subgroup with high physical and/or psychological symptoms, one subgroup with low physical and psychological symptoms, and one or more moderate symptom subgroup(s). The presence of distinct symptom subgroups argues against uniform assessment and treatment of fibromyalgia and supports the need for further research to identify symptom patterns and enhance symptom management.
Existing research examining the longitudinal stability of individual fibromyalgia symptoms suggests that changes in individual symptoms in patients with fibromyalgia are generally small and of doubtful clinical significance [
8]. No study to date has examined changes in a patient’s entire symptom profile over time. Therefore, the purpose of this study was to examine the stability of individual patient profiles (symptom clusters) over a 2-year period in a well-characterized cohort of patients with fibromyalgia.
Results
From the original sample of 581 patients, 433 (74.5%) completed the follow-up survey and had complete data on all clustering variables. Demographic characteristics of the sample with complete data are shown in Table
1. The median age was 56 years, and the median body mass index was 29 kg/m
2. The median total FIQR score was 57, and the majority of patients (83.8%;
n = 363) continued to meet Fibromyalgia Research Survey criteria for the diagnosis of fibromyalgia at follow-up.
Table 1
Patient characteristics at baseline (N = 433)
Age, years | 56 (20–85) |
BMI, kg/m2 | 29 (13–58) |
BMI category | (n = 431) |
< 25 kg/m2 | 114 (26) |
25–29.99 kg/m2 | 125 (29) |
30–34.99 kg/m2 | 98 (23) |
≥ 35 kg/m2 | 94 (22) |
FIQR total score | 57 (8–96) |
SF-36 physical composite score | 30 (8–53) |
SF-36 mental composite score | 41 (7–67) |
FM Research Survey Criteria, WPI | 13 (3–19) |
FM Research Survey Criteria, SS | 9 (5–12) |
Concordance between cluster membership at baseline and follow-up is shown in Table
2. Overall, 251 participants (58%) remained in their baseline cluster at follow-up, for a κ statistic of 0.44 (95% CI 0.37–0.50), which suggests moderate agreement or stability between baseline and 2-year follow-up. The highest concordance was observed for participants originally classified in cluster 1 and cluster 4, for which 68% and 70%, respectively, remained in the same cluster at follow-up. Baseline cluster 2 and 3 patients showed more reclassification, with only 53% and 43%, respectively, in the same cluster at follow-up. Baseline cluster 2 patients were most commonly reclassified into cluster 1 (21%), followed by cluster 3 (19%). Baseline cluster 3 patients were also most commonly reclassified into cluster 1 (32%). Patients at the extremes of severity (cluster 1 low severity, and cluster 4 high severity) were rarely reclassified in the opposite extreme at follow-up: three baseline cluster 1 patients (3%) and two baseline cluster 4 patients (2%).
Table 2
Agreement between baseline cluster assignment and follow-up cluster assignment
1 |
79 (68)
| 21 (18) | 13 (11) | 3 (3) | 116 |
2 | 27 (21) |
69 (53)
| 25 (19) | 10 (8) | 131 |
3 | 32 (32) | 11 (11) |
43 (43)
| 14 (14) | 100 |
4 | 2 (2) | 9 (10) | 15 (17) |
60 (70)
| 86 |
n
| 140 | 110 | 96 | 87 | 433 |
When evaluating the stability of individual symptoms, intraclass correlation coefficient values showed generally good consistency between the two time points for individual symptom domain measures, with values of 0.60 or greater for all scales except FIQR stiffness, which had an intraclass correlation coefficient of only 0.23 (Table
3). Small but statistically significant improvements were observed at follow-up for the symptoms of fatigue, stiffness, and dyscognition. The FIQR total score and SF-36 physical composite score also improved significantly by means of 1.50 points (
P = 0.02) and 0.76 points (
P = 0.03), respectively. Significant worsening was not observed for any scale.
Table 3
Consistency and degree of change between baseline and follow-up for individual measures
Cluster analysis measures |
MFI | 0.61 | 15.7 (3.6) | 15.2 (3.7) | −0.54 (−0.85, −0.23) | 44 (10) | 346 (80) | 43 (10) |
MOS-Sleep | 0.66 | 55.2 (18.9) | 55.4 (19.1) | 0.20 (−1.29, 1.69) | 77 (18) | 304 (70) | 52 (12) |
BPI severity | 0.62 | 5.1 (1.8) | 5.0 (1.8) | −0.11 (−0.26, 0.03) | 79 (18) | 285 (66) | 69 (16) |
FIQR function | 0.72 | 14.0 (7.2) | 13.7 (7.4) | −0.30 (−0.82, 0.22) | 106 (24) | 226 (52) | 97 (22) |
FIQR stiffness | 0.23 | 7.2 (2.2) | 6.6 (2.4) | −0.62 (−0.85, −0.40) | 53 (12) | 287 (66) | 90 (21) |
MASQ | 0.80 | 95.1 (21.8) | 93.7 (22.3) | −1.4 (−2.7, −0.05) | 19 (4) | 405 (94) | 9 (2) |
POMS depression-dejection | 0.61 | 6.6 (5.0) | 6.3 (4.9) | −0.25 (−0.67, 0.16) | 128 (30) | 172 (40) | 133 (31) |
POMS tension-anxiety | 0.60 | 7.0 (4.6) | 7.2 (4.7) | 0.15 (−0.24, 0.55) | 131 (30) | 198 (46) | 104 (24) |
Global measures not included in cluster analysis |
FIQR total (n = 410) | 0.75 | 55.4 (19.0) | 53.9 (19.9) | −1.50 (−2.84, −0.17) | 59 (14) | 300 (73) | 51 (12) |
SF-36 physical (n = 380) | 0.69 | 30.3 (8.7) | 31.0 (8.6) | 0.76 (0.07, 1.44) | 19 (5) | 305 (80) | 56 (15) |
SF-36 mental (n = 380) | 0.60 | 40.6 (12.3) | 40.2 (12.7) | −0.37 (−1.50, 0.75) | 47 (12) | 267 (70) | 66 (17) |
FM Research Survey Criteria, WPI | 0.47 | 12.6 (3.8) | 12.1 (4.0) | −0.58 (−0.20, −0.95) | 75 (17) | 272 (63) | 86 (20) |
FM Research Survey Criteria, SS | 0.57 | 8.8 (2.0) | 8.2 (2.3) | −0.63 (−0.43, −0.81) | 31 (7) | 344 (80) | 57 (13) |
Although statistically significant changes between the two time points were observed for some instruments, there were few clinically meaningful differences, with most patients remaining within 30% of baseline at follow-up for all instruments except POMS depression-dejection and POMS tension-anxiety (Table
3). On the depression-dejection and tension-anxiety scales, more than half of patients changed by more than 30% from baseline to follow-up, but those changes reflected both improvements (31% and 24%, respectively, improving by ≥ 30%) and worsening (30% worsening by ≥ 30% for each instrument). On global measures, 70% or more of patients remained within 30% of baseline: 73% for FIQR, 80% for SF-36 physical, and 70% for SF-36 mental.
For patients who moved to a better cluster (i.e., a lower symptom burden profile) between baseline and follow-up, the symptom domains demonstrating the largest changes were depression (mean 26% decrease from baseline), stiffness (mean 23% decrease from baseline), and function (mean 21% decrease from baseline) (Table
4). Statistically significant changes were observed for all symptoms except sleep. Similarly, for patients who moved to a worse cluster, all symptoms showed a statistically significant increase. Among those who were in a worse cluster at follow-up, the symptom domains demonstrating the largest changes were depression (mean 112% increase from baseline), anxiety (mean 102% increase from baseline), and function (mean 58% increase in score, indicating decreasing function, from baseline).
Table 4
Percentage change on cluster analysis measures: movement to a better cluster or worse clustera
MFI | −12.42 (−16.96, −7.89) | 9.92 (4.73, 15.10) |
MOS-Sleep | −3.72 (−12.57, 5.13) | 25.46 (13.36, 37.56) |
BPI | −15.62 (−21.81, −9.44) | 35.28 (16.01, 54.54) |
FIQR function | −21.36 (−30.70, −12.02) | 58.38 (24.45, 92.30) |
FIQR stiffness | −23.26 (−32.24, −14.28) | 31.27 (14.03, 48.52) |
MASQ | −5.85 (−8.28, −3.43) | 8.59 (5.07, 12.10) |
POMS depression-dejection | −26.28 (−39.67, −12.88) | 111.95 (73.43, 150.48) |
POMS tension-anxiety | −15.97 (−29.11, −2.83) | 101.90 (62.73, 141.06) |
Discussion
The results of our study indicate relative stability in fibromyalgia symptom severity and symptom clusters from baseline to 2-year follow-up. This cluster stability was most apparent in two of the four subgroups: those with the lowest level of symptoms (cluster 1) and those with the highest level of symptoms (cluster 4). In contrast, patients in clusters 2 and 3 with moderate symptom severity demonstrated greater fluctuation. Patients who shifted to a different cluster at follow-up were slightly more likely to move to a lower severity cluster than to a higher one.
Although the clinical implications of these findings are not fully understood, our results suggest that most patients with fibromyalgia generally did not have progressive worsening and maintained their baseline symptom severity profile. This information provides the opportunity for further research evaluating individualized management tailored to each patient’s multisymptom profile. Fibromyalgia management remains challenging for clinicians despite the availability of several pharmacologic and nonpharmacologic therapies [
40]. This challenge may be partly due to the heterogeneity of the syndrome and a one-size-fits-all approach that is currently used for disease management in patients of varying symptom severity patterns. Improved methods of clinical stratification to categorize symptom severity across all relevant domains at the time of diagnosis could improve clinical management of fibromyalgia and warrants further study.
Patients in the most severe cluster (cluster 4) and the lowest symptom severity cluster (cluster 1) showed the greatest stability, with 70% and 68%, respectively, remaining in the same cluster at follow-up. Given their overall low symptom severity, cluster 1 patients may be excellent candidates for nonpharmacologic treatments (e.g., aerobic and strength training, cognitive behavioral therapy) as currently recommended by the European League Against Rheumatism [
41]. In contrast, cluster 4 patients, whose symptoms remained severe over time for the most part, may be best served by early referral to a multidisciplinary pain clinic as recommended by the European League Against Rheumatism [
41], rather than treating patients in the primary care setting or referring them to multiple subspecialists. In addition, patients with a high degree of anxiety and depression may also benefit from early referral to mental health specialists.
Of interest, among patients who moved to a worse cluster, the symptoms that drove the change were predominantly depression and anxiety. This is not surprising since previous studies report the detrimental effects of comorbid depression and anxiety on overall symptom severity [
42,
43] and because of the bidirectional effect of depression in samples of patients with fibromyalgia [
44]. Research has also demonstrated that patients with fibromyalgia with comorbid mood disorders and anxiety tend to have poorer outcomes than patients with predominantly physical and lower psychological symptoms [
42]. Our finding that mood and anxiety tended to drive changes in cluster categorization highlights the importance of regular mood and anxiety assessment and referral to mental health specialists, as appropriate, in patients with fibromyalgia who have worsening symptoms.
Our approach using multiple questionnaires is not suited for routine clinical care. It has been suggested that the stratification of fibromyalgia severity in busy clinical practices could be accomplished through the use of single instruments [
45]. For example, the visual analogue scales of the FIQR, the complete FIQR (although this involves more complex scoring), or the Fibromyalgia Symptom Score could be used to rapidly gauge both severity and constellation of fibromyalgia symptoms [
45].
The current study has several limitations. One key limitation was the lack of data regarding medications and other treatment modalities that patients may have used during the 2-year time frame; however, it was not feasible to collect information at the level of detail required, and patients’ recall of medications and changes in medication regimen are frequently inaccurate [
46]. A second limitation is the loss of approximately 25% of our sample at follow-up, which may have biased our results. However, patients without follow-up did not differ significantly in baseline cluster distribution from those who did complete follow-up (data not shown). Therefore, it is unlikely that this influenced our results. Another major limitation is the lack of data on work/disability status, socioeconomic status, and medical and psychiatric comorbid conditions. These data were not collected as part of this mailed survey to limit the participant burden associated with the inclusion of more than 200 questionnaire items. In our previous cross-sectional study [
2], the percentage of patients on work disability differed significantly between clusters, with the highest percentage of patients on disability in cluster 4 and the lowest percentage in cluster 1. In addition, because of our inclusion of only women with fibromyalgia identified through a clinical registry [
47], we cannot comment on the generalizability of our results to community samples, men, or adolescents with fibromyalgia.