To determine if assigning the label of fibromyalgia (FM) to individuals with chronic widespread pain has a significant effect on long-term health status, function, and health service utilization.
To determine if assigning the label of fibromyalgia (FM) to individuals with chronic widespread pain has a significant effect on long-term health status, function, and health service utilization.
In the London Fibromyalgia Epidemiology Study, 100 individuals with FM were identified by screening 3,395 non-institutionalized adults. Only 28 of the 100 had been previously diagnosed with FM; for 72, the diagnostic label was new. All 28 with prediagnosed FM were female compared with 58 of the 72 newly diagnosed cases. In a prospective, within-group comparison, we compared previously non-labeled FM cases at study entry (prelabeling) and at 18 and 36 months followup (postlabeling) with respect to general health status, fibromyalgia-related symptoms, and all items from the Fibromyalgia Impact Questionnaire (FIQ) (including total FIQ score, and several measures of health service utilization) to see if health status, function, and health services utilization had changed, using paired t-tests. We also compared percentage reporting work disability at baseline and 18 months using Pearson's χ2.
Fifty-six (78%) of the original 72 newly diagnosed FM cases were available for reassessment at 18 months, and 43 (60%) at 36 months. Although physical functioning decreased slightly over time, there also was a statistically significant improvement in satisfaction with health, and newly diagnosed FM cases reported fewer symptoms and major symptoms over the long term. No other differences in clinical status or health service use occurred over time.
The FM label does not have a meaningful adverse affect on clinical outcome over the long term. Further research is necessary to document the short- and long-term effect of labeling in the chronic pain patient.
In 1990, seemingly sensitive and specific classification criteria for fibromyalgia (FM) were published following a multicenter study of FM patients versus controls with other rheumatic disorders (1). These criteria were developed as an attempt to categorize a subset of rheumatology patients exhibiting chronic widespread pain and several other symptoms. Although primarily designed for research purposes, these criteria now are widely used to diagnose patients as having FM. For a variety of reasons, the labeling of certain chronic pain patients as having FM has sparked much debate.
One of the central arguments in this debate is that the label of fibromyalgia, in itself, might precipitate or exacerbate behavior that has been variably termed “illness behavior,” “learned pain” (2) and learned helplessness (3). This potentially would result in heightened symptoms, worsened function, increased disability claims, and increased health care seeking behavior (4–6). A case for the danger of labeling has been made in black lung disease as it occurred in coal mining areas in the US. In this example, large numbers of miners were labeled with black lung disease and granted disability pensions based on radiographic findings, despite entirely normal pulmonary function (7). To date, however, there is no good evidence that the FM label has either a deleterious or beneficial effect.
The current study examines the long-term implications of assigning the label of FM to an individual with chronic widespread pain. The study was performed prospectively in a representative community cohort of adults with chronic widespread pain. We present the results of a between-group comparison between the 72 previously undiagnosed and the 28 previously diagnosed FM cases. We then present the results of within-group (paired) comparisons of the 72 newly diagnosed individuals at 18 months and 36 months postlabeling versus prelabeling (baseline) to determine if labeling had a detrimental or any other effect. Our specific objectives were to compare previously diagnosed and previously undiagnosed FM cases at baseline with respect to health status, function, disability status, and health services utilization; and to then prospectively determine the effect of the FM label on future health status, function, disability status, and health services utilization. To do so, we compared 72 newly diagnosed, and hence newly-labeled cases of FM at 18 and 36 months followup (postlabeling) versus baseline (prelabeling). Specifically, we assessed the following over time: mean global health status; mean number of symptoms and major symptoms on a 41-item symptoms checklist; mean global fibromyalgia activity; mean levels of pain, fatigue, morning fatigue, stiffness, anxiety, and depression on 100 mm visual analog scales; mean visits to general practitioners, specialists and other health care providers over the preceding 2 weeks and one year; mean number of prescription and total drugs currently used; and percentage of subjects claiming total disability.
One hundred community-dwelling adults with FM were identified in a random digit dialing, cross-sectional telephone survey of non-institutionalized adults in London, Ontario, a community of 341,320 persons (1991 census) in Southwestern Ontario. The survey method and the screening instrument have been described in detail elsewhere (8, 9). All 100 FM cases screened positive for chronic widespread musculoskeletal pain and were subsequently examined by a rheumatologist to confirm that they met the full 1990 American College of Rheumatology (ACR) classification criteria for FM: 1) reporting widespread (axial and peripheral, right and left, and above and below the waist) musculoskeletal pain of no less than 3 months duration, and 2) reporting pain upon digital palpation of no fewer than 11 of the 18 ACR specified tender points.
After this examination, the 100 FM subjects (FC) and 76 pain controls (PC) completed a more detailed health questionnaire. Among the 100 FM subjects, 72 denied previously having been told they had FM; 58 of these 72 were female (81%) compared with all 28 prediagnosed cases being female. All 100 FC and 76 PC were invited to participate in a prospective 5-year study, with reassessments at 18, 36, and 60 months, to determine the natural history of FM in a community cohort.
Each reassessment consisted of an interview and examination by a rheumatologist to determine if a subject continued to meet the case definition of FM, followed by completion of a detailed health questionnaire, which included:
For between-group and within-group comparisons, Student's t-tests and paired Student's t-tests were used, respectively, for all continuous variables. For nominal and ordinal data, Pearson's χ2 was used. All tests were two-tailed. Because type II error was of greater concern than type I error, we attempted to maximize the likelihood of finding between- and within-group differences by not adjusting for multiple comparisons. Because only a small minority of subjects were males, all analyses were performed both including and excluding males; because the 2 methods did not produce any discrepancies, data are presented for the sexes combined. To address the concern of loss-to-followup, we compared baseline characteristics of participants and nonparticipants at 18 and 36 months.
All 28 previously diagnosed FM cases were female, compared to 58 of 72 (81%) previously undiagnosed cases (95% confidence intervals [95% CI] 9.9–28.9) (Table 1). Mean ages for previously diagnosed and undiagnosed cases were 45.8 and 48.6 years, respectively (−9.2, 3.4). Previously diagnosed cases were more dissatisfied with their health. They reported more symptoms (26.3 versus 21.2; 95% CI 1.6–8.4) and more major symptoms (11.9 versus 8.0; 95% CI 1.0–6.6) than previously undiagnosed cases. They had more FM tender points (15.8 versus 13.8; 95% CI 1.0–3.0). There was a trend for previously diagnosed cases to have increased severity of every other clinical variable, but no differences achieved statistical significance. There was, however, no clear trend with respect to health services use. More than half (53.6%) of the 28 previously-diagnosed FM cases claimed being totally disabled, compared to only 22.2% of the 72 previously undiagnosed cases (95% CI 3.3–59.5).
|Variable||Previously diagnosed n = 28||Previously undiagnosed n = 72||95% confidence interval of difference|
|Mean age (years)||45.8||48.6||(−9.2, 3.4)|
|% female||100.0||80.6||(9.9, 28.9)*|
|Global ill health||4.0||3.7||(−0.2, 0.6)|
|Dissatisfaction with health||3.3||2.8||(0.2, 0.8)*|
|Number of symptoms||26.3||21.2||(1.6, 8.4)*|
|Number of major symptoms||11.9||8.0||(1.0, 6.6)*|
|Number of tender points||15.8||13.8||(1.0, 3.0)*|
|FIQ total score||68.3||63.4||(−1.6, 11.5)|
|FIQ physical score||51.3||40.7||(−0.4, 21.6)|
|VAS pain||72.6||63.8||(−0.2, 17.7)|
|VAS fatigue||79.4||75.5||(−4.1, 11.9)|
|VAS a.m. fatigue||76.3||71.7||(−4.9, 14.1)|
|VAS stiffness||75.7||67.7||(−0.8, 16.9)|
|VAS anxiety||65.3||56.0||(−2.7, 21.3)|
|VAS depression||56.4||53.2||(−10.2, 16.5)|
|Annual visits to GP||11.6||8.4||(−2.8, 9.2)|
|Annual visits to specialist MD||2.6||2.9||(−3.2, 2.6)|
|Annual visits to emergency room||0.6||1.1||(−1.6, 0.6)|
|Hospital inpatient days||2.1||2.0||(−3.4, 3.7)|
|Number of prescription drugs||3.3||2.7||(−0.5, 1.7)|
|% claiming total disability||53.6||22.2||(3.3, 59.5)*|
|% receiving disability pensions||39.3||20.8||(−9.0, 46.0)|
Sixteen of 72 (22.2%) newly-diagnosed FM cases had dropped out of the study by the 18-month followup assessment, and 29 (40.3%) had dropped out by 36 months. Nonparticipants were disproportionately male; whereas less than 10% of participants were male at 18 and 36 months, the percentages of nonparticipants who were male were 56.2% and 34.5%, respectively. Otherwise, participants and nonparticipants were similar (Table 2). At 18 months, there did not even appear to be a trend towards one group being clinically more active at baseline, with 4 of 11 clinical variables being worse for nonparticipants, 6 being worse for participants, and one being equal. At 36 months, there was a trend towards participants having been worse at baseline, with 10 of 11 baseline variables being worse in this group; all these differences were slight, however, and none achieved statistical significance.
|At 18 months||At 36 months|
|Participants n = 56||Lost to followup n = 16||95% CI||Participants n = 43||Lost to followup n = 29||95% CI|
|Global ill health||3.7||3.8||(−0.5–0.5)||3.7||3.7||(−0.4–0.4)|
|Dissatisfaction with health||2.9||2.6||(−0.1–0.7)||3.0||2.6||(−0.01–0.7)|
|Number of symptoms||21.2||21.2||(−4.3–4.4)||21.9||20.1||(−1.8–5.4)|
|Number of major symptoms||8.3||7.1||(−2.1–4.5)||8.5||7.4||(−1.7–3.9)|
|Number of tender points||14.0||13.3||(−0.5–2.0)||14.0||13.6||(−0.7–1.5)|
|FIQ total score||64.5||58.9||(−3.8–15.0)||65.5||59.4||(−1.7–13.9)|
|FIQ physical score||40.2||42.8||(−18.3–13.1)||43.1||36.5||(−6.6–19.9)|
|Annual visits to GP||8.5||7.8||(−5.2–6.8)||8.8||7.8||(−4.1–6.1)|
|Annual visits to specialist MD||2.6||3.9||(−5.4–2.7)||3.0||2.7||(−3.1–3.8)|
|% claiming total disability||21.4||25.0||(−35.6–28.4)||23.3||20.7||(−24.8–30.0)|
|% receiving disability pensions||19.6||25.0||(−37.0–26.2)||20.9||20.7||(−26.7–27.1)|
Fifty-one of the 56 participants (91%) at the 18-month assessment were female (Table 3). Again at 36 months, 91% (39 of 43) of participants were female. By 18 months, these newly labeled FM cases reported more symptoms (23.4 at 18 months versus 21.4 at baseline; 95% CI 0.3–3.7) and more major symptoms (9.6 versus 8.1; 95% CI 0.5–2.9). However, by 36 months, the same two variables were improved versus baseline; subjects reported almost 6 fewer symptoms (15.4 versus 21.2; 95% CI −8.6, −3.2) and fewer major symptoms (6.6 versus 8.3; 95% CI −0.2, −3.2). By 36 months, the newly diagnosed FM cases reported a clinically and statistically significant decrease in dissatisfaction with health (2.2 versus 3.0 on a 5-point Likert scale; 95% CI −1.2, −0.4), but also reported somewhat greater functional limitations, as measured by the first item of the FIQ (percentage dysfunction = 50.9 versus 43.1; 95% CI 1.8–13.5).
|Variable||Baseline||18 months||95% CI||Baseline||36 months||95% CI|
|(n = 56; 51 female)||(n = 43; 39 female)|
|Global ill health||3.7||3.7||(−0.3–0.3)||3.7||4.0||(−0.5–0.02)|
|Dissatisfaction with health||2.9||2.8||(−0.1–0.3)||3.0||2.2||(−1.2–−0.4)*|
|Number of symptoms||21.4||23.4||(0.3–3.7)*||21.2||15.4||(−8.4–−3.2)*|
|Number of major symptoms||8.1||9.6||(0.5–2.9)*||8.3||6.6||(−0.2–−3.2)*|
|Number of tender points||14.0||14.4||(−0.4–1.2)||14.0||14.7||(−0.4–1.7)|
|FIQ total score||65.2||63.3||(−5.6–1.9)||65.5||64.2||(−6.0–3.4)|
|FIQ physical score||40.2||42.6||(−3.8–8.7)||43.1||50.9||(1.8–13.5)*|
|VAS a.m. fatigue||74.5||71.8||(−9.1–3.9)||75.7||74.6||(−8.5–6.2)|
|Annual visits to GP||8.5||5.9||(−5.5–0.2)||8.7||6.0||(−6.2–0.7)|
|Annual visits to specialist MD||2.6||2.7||(−1.2–1.0)||3.0||1.8||(−0.5–2.9)|
|Annual visits to emergency room||0.7||0.7||(−0.4–0.5)||0.5||0.6||(−0.3–0.4)|
|Hospital inpatient days||1.7||2.6||(−2.6–4.4)||2.3||1.3||(−3.9–1.9)|
|Number of prescription drugs||2.9||2.7||(−0.9–0.4)||2.8||2.9||(−0.5–0.8)|
|% claiming total disability||21.4||23.2||(−23.6–20.0)||23.3||34.9||(−38.5–15.3)|
|% receiving disability pensions||19.6||14.3||(−14.3–24.9)||20.9||30.2||(−35.2–16.6)|
Although falling far short of being statistically significant, the percentage of FM patients who reported being totally disabled (34.9% from 23.3%) and who reported receiving disability checks (30.2% from 20.9%) increased by approximately half. Otherwise, there did not appear to be any trend toward clinical worsening over time. Of twenty clinical variables studied at 18 months, 9 were at least somewhat improved (even if there was no statistically significant change), 8 were worse, and 3 remained unchanged. At 36 months, 12 were improved and 8 worse versus baseline (only 4 variables were statistically different).
To our knowledge, our study is the first published attempt to assess the effect of the diagnostic label of fibromyalgia on clinical course and health services utilization. Our data suggest that the FM label does not have a meaningful adverse effect on clinical outcome over the long term.
Our study has two significant strengths. First, it is prospective. Second, it involves a representative community cohort of adults with chronic widespread pain, individuals who were not necessarily seeking a diagnosis for their symptoms. We observed that, although those carrying a pre-existent diagnosis of FM at baseline tended to be clinically worse than those without the FM label, those to whom we assigned the FM label did not worsen over time. What this suggests is that, although those who are clinically worse are more likely to be diagnosed with FM, the FM label itself does not adversely affect their clinical course. In fact, although patients appear to become somewhat less active over time, their satisfaction with health improves, and their number of symptoms and major symptoms lessen over the long term. It is conceivable that the FM label allows for more appropriate medical management (which might include pacing of activities and hence less activity) to account for these positive changes.
Our findings are in direct contrast to many opinions recently published in the medical literature, in which certain authors suggest abolishing the FM label for something “more neutral” (12–15). These authors do not offer scientific evidence supporting their claims. Some report that the FM label has resulted in an epidemic of FM (15, 16). However, since prevalence studies have never been repeated within any given general population, there are no data to support the assumption that prevalence is increasing. It certainly is possible that the apparent increase in prevalence within clinical practice is merely an artifact of increased recognition by health care practitioners now that a case definition for FM has been published and disseminated.
Other authors present detailed anecdotal accounts of patients who, once labeled, “are drawn into medicalization and dependency” (17) and “become victims worthy of a star appearance on the Oprah Winfrey show” (18). Our data do not refute the possibility that a minority of patients labeled with fibromyalgia become obsessed with their illness, leading to increased illness behavior. However, the same might be said of certain patients labeled with hypertension or coronary artery disease. Moreover, our data argue against there being any net adverse affect of the FM label on adults with chronic widespread pain who meet the ACR criteria.
The reasons for the apparently erroneous opinion that the FM label itself creates illness likely are numerous. One possible reason is referral bias. Fred Wolfe has described the “fibromyalgia funnel,” a series of sieves through which potential study subjects must pass as they proceed from being sick at home (hence not part of a rheumatology practice or FM research study) to being sick within a clinical trial or other research study (18). Invariably, those who end up in tertiary clinics and clinical trials have been sick longer, have had worse symptoms, and/or have coped less well with their symptoms than those with the same illness not being followed by a rheumatologist. In fact, several studies have demonstrated that not just FM, but well-accepted disorders such as rheumatoid arthritis are clinically less severe in community versus clinic samples (19). Our study supports the FM funnel assumption—the 28 previously diagnosed FM cases were clinically worse than those not previously given the FM label. In other words, those who were worse at study entry had been investigated to the point of confirming the FM diagnosis; those with less severe symptoms had not.
Our study essentially was able to avoid the issue of referral bias because of the probabilistic sample we used, and because subjects were selected from the general community, rather than from clinic or hospital populations. One potential criticism of our study is that we made no attempt to control for subsequent treatment that labeled individuals might have pursued following their diagnosis. However, such criticism is unfounded, since seeking disease-specific treatment after receiving a diagnosis must be considered an acceptable, and usually desirable, outcome. Moreover, our data suggest that individuals labeled with fibromyalgia do not increase their utilization of health services.
Another potential criticism is that our study lacked sufficient power to detect the within group differences. However, there did not even appear to be a trend towards worsening, with as many variables non-statistically improved as non-statistically worsened over time.
It may be that, in certain settings such as the medicolegal context, the FM label may contribute to heightened illness behavior and hence, might be problematic (16), but such an assumption is unproved. We did not specifically address this issue. However, although the percentage of our FM cases claiming total disability did increase by 50% (from 23% to 35%) after labeling (not statistically significant), two-thirds of FM cases failed to consider themselves disabled over a 3-year period. This suggests that the FM label did not result in a rush of disability claims. Our FM labeled subjects did function at a slightly lower level after 3 years, but also reported fewer symptoms, fewer major symptoms, and less dissatisfaction with health. It is entirely conceivable that the FM label resulted in FM-specific treatment which would include advice regarding pacing of activities, which in turn might account for the apparent slight decrease in function.
There are numerous inherent problems that exist when developing and utilizing case definitions for symptom-based conditions, largely related to specificity (20, 21). However, our data suggest that making patients worse is not a significant problem with the FM label. Further research is needed to document the trajectory and implications of the apparent small decline in physical function noted in our sample.