Opinion on Preliminary Guidelines for the Clinical Diagnostic Criteria for Fibromyalgia
The 1990 American College of Rheumatology Classification Criteria for fibromyalgia have been the gold-standard criteria for diagnosing this disorder for the last 20 years.1 These criteria classify fibromyalgia as a disorder characterized by widespread pain and undue tenderness to moderate palpation pressure in at least 11 out of 18 specified locations (tender points). In general, classification criteria are strictly intended for use in epidemiological studies and not necessarily for clinical diagnosis. However the 1990 paper did note, “the sensitivity of the criteria suggests that they may be useful for diagnosis as well as classification.” Over the ensuing 20 years, it has become apparent that some patients with typical FM symptoms do not meet the criteria of tender points, yet no other diagnosis is apparent on follow up. In some other cases, patients may experience worthwhile improvement to the extent that they no longer meet the tender point criteria. Do they now suddenly cease to have FM? Lastly, the clinical evaluation of tender points is an acquired skill that is not routinely taught in medical school and is thus seldom or improperly used by primary care physicians.2
Importantly, the 1990 criteria do not embrace the many other symptoms that are commonly reported by FM patients and hence they define a disorder that is based on the experience of pain alone.3 It is noteworthy that many other common FM symptoms—such as fatigue, poor sleep, cold sensitivity, etc.—were evaluated in the development of the 1990 criteria and found not to enhance the diagnosis. In this respect the discussion in the 1990 paper noted: “Of particular interest was the finding that the simultaneous occurrence of sleep disturbance, fatigue, and morning stiffness—required in certain previous criteria sets—was found in only 56% of patients…no combination or set of combinations of tender points and symptoms could be found that performed better than the tender point and widespread pain criteria.” Thus the 1990 criteria did not use these other common manifestations of FM in its final recommendations, based on strictly statistical reasons relating to sensitivity and specificity. Bearing these critiques in mind, there has been an increasing awareness that the 1990 criteria needed revising, especially when it comes to making a diagnosis of FM in the clinical setting.3
Herein, I discuss a recent article that attempts to provide alternative diagnostic criteria that also includes a measurement of symptom severity.4 These new preliminary criteria were based on an analysis data from previously diagnosed FM patients (either by ACR criteria or the physicians’ “clinical” impression) and patients with non-inflammatory pain disorders, such as osteoarthritis, bursitis, tendinitis, neck pain and back pain. All subjects were culled from the practices of 32 “interested” rheumatologists. The data analysis was based on a 2-stage design. The study that led to this publication was funded by Lilly Research Laboratories but this sponsor did not participate in the design or analysis of the study, nor did they review the manuscript.
All subjects (n=610) completed the following five questionnaires before seeing the physician:
- Whether they had pain in any of 19 locations (the Widespread Pain Index or WPI)
- Rating of four symptoms (pain, fatigue, ability to sleep and un-refreshing sleep) on a 0 scale
- The Health Assessment Question-naire II functional disability scale
- The number of medications used
- How many of 56 designated symptoms they experienced
All physicians (n=32) independently (i.e. without looking at the patients’ responses) completed the WPI, the Symptom Severity Scale (SS), and also performed an ACR-refined tender point evaluation. The SS was composed of 3 questions (fatigue, cognitive problems, waking un-refreshed and extent of somatic symptoms) rated on a scale of 0-3; thus it had a range of 0-12. Th extent of somatic symptoms was reduced to a 0-3 scale from a list of 41 symptoms.
They also noted if the subjects had any of the following problems: muscle pain, muscle weakness, irritable bowel syndrome, fatigue, cognitive problems, headaches, abdominal pains/cramps, paresthesias, dizziness, sleep problems, depression, anxiety, constipation, diarrhea, interstitial cystitis and muscle tenderness. Lastly, they indicated their certainty of the prior diagnosis on a 0-10 scale (0=very uncertain, 10= very certain).
This stage was completed only by twenty-two physicians (the ten FM experts did not take part in this stage). The patients were excluded from this stage. From the same WPI list used by the patients from the first stage, they rated the pain extent as 0-3, 4-6, 7-10 or ≥ 11. They also performed:
- The ACR tender point evaluation
- Recorded the presence/absence of muscle pain
- Recorded the presence/absence of muscle tenderness
- Recorded the presence/absence of irritable bowel syndrome
- Provided a rating of somatic symptoms from a list of 41 symptoms (scaled as:0=few or no symptoms, 1=mild number of symptoms, 2= moderate number of symptoms, 3=great number of symptoms)
- Recorded the 3 categorical scales of: 1) un-refreshing sleep, 2) cognitive problems, and 3) fatigue. All symptoms are scored on a 0-3 scale: (scaled 0=no problem, 1=mild problem, 2=moderate problem, and 3=severe problem).
The phase 2 study data were used to determine if a shortened questionnaire would work as well in categorizing fibromyalgia as the longer, more detailed phase 1 assessments.
Data from the first stage was analyzed from three viewpoints:
- A short set that included WPI and the categorical scores for pain, fatigue, sleep disturbance, mood, cognitive problems, un-refreshing sleep and somatic symptoms.
- An intermediate set that included all the variables in the short set plus all of the individual somatic symptoms.
- The full set included all of the study variables.
Based on a series of standard statistical tests (t tests, regression analysis, correlation analyses, classification tree analyses) as well as some more “sophisticated” analyses (“out of bag” error results, random forest analyses, Gini index, recursive partitioning R analysis); these data were presented as a probability density function; the author advises the reader to think of this as a “smoothed-out” version of a histogram.
Some 258 valid FM patients and 256 control patients were entered into the final analysis. Of the 258 FM subjects, 63.6% had an ACR-based diagnosis and the rest had a symptom-based diagnosis. All subjects were categorized into three groups based on prior diagnosis and ACR classification: