Objective: A randomized controlled trial was conducted to assess reliability and accuracy in identification of fibromyalgia (FM), motivated simulation, and normal controls.
Methods: Eight female subjects with chronic FM were age matched with 19 healthy female volunteers. The volunteers were randomized to a financially motivated "simulator" group who were paid to simulate FM, or to a "normal control" group. Examiners under blinded conditions rated tender and control points, and illness behavior. Intraclass correlation coefficients and F values showed that counts of tender points significantly discriminated the 3 groups. Variance was mostly due to experimental groups and not to observer or error factors. In this study, simulators could not be discriminated from normals or FM subjects on the basis of tenderness at "control points." Examiner ratings of illness behavior (UAB), and subjects' self-ratings for pain showed that FM subjects had the highest scores, normals the lowest, and simulators had mean scores midway between the mean FM and simulator. On grip strength, the normals obtained the highest scores, the simulators the lowest, and the FM subjects had scores midway between those of the other 2 groups. Diagnostic accuracy of the blinded examiners in distinguishing FM from simulators and from normals was 80%, and for correct diagnosis the kappa value was significant at 0.69. Despite this, simulators were misidentified as FM in 1/3 of judgments, and FM was misidentified as simulators in 1/5 of judgments.
Conclusion: Under randomized blinded conditions, examiners using the American College of Rheumatology criteria for FM and other bedside observations are able to distinguish chronic FM, normal individuals, and motivated simulators with 80% accuracy, with a good level of agreement and reliability in tender point counts. Our results do not provide a "test for malingering," and it is likely that an important minority of motivated simulators and of FM subjects will be misidentified.