The primary objectives of this study were to 1) conduct confirmatory analyses of the performance of a new patient- and physician-completed fibromyalgia (FM) screener, the Arnold Fibromyalgia Diagnostic Screen (AFDS), using the 1990 American College of Rheumatology (ACR) diagnostic criteria as the gold standard comparison and 2) select the AFDS scoring model that maximizes accuracy in predicting an ACR FM diagnosis. Secondary objectives included 1) evaluating the screening ability of the London Fibromyalgia Epidemiology Study Screening Questionnaire (LFESSQ) and the ACR Diagnostic Criteria (ACR-FDC) and 2) documenting time to completion and participant preference among the AFDS, LFESSQ, and the ACR-FDC.
Method: 150 adult participants with chronic pain, half of whom had received a physician diagnosis of FM, were enrolled in this multicenter cross-sectional study. The study visit included a physician-conducted evaluation to capture the clinician-reported components of the AFDS, assessment of whether the patient met the ACR criteria for FM, and assessment of whether the patient had FM based on clinical judgment. Additionally, the study visit included patient completion of the three screening questionnaires and venipuncture. The sensitivity and specificity for an FM diagnosis based on the ACR criteria was tabulated for each of six AFDS scoring models. Kappa, Youden’s index, overall accuracy, and a likelihood ratio were also calculated. Chi-square contingency tables with odds ratios were created predicting the dependent variable (0 = no ACR FM diagnosis, 1 = ACR FM diagnosis) for two of the AFDS models (the primary scoring model and the scoring model with the highest Youden’s index) as well as for the LFESSQ and ACR-FDC. The three patient preference items were summarized using counts and percentages; means, medians, and standard deviations (SD) were calculated for the participant time to complete each questionnaire. All analyses were conducted using SAS. All statistical tests were 2-tailed. A conservative type 1 error rate of 1% (α = 0.01) was applied to each individual hypothesis test.
Result: Item-level analyses provided support for the response categories and the predictive ability of most of the individual AFDS items. Additionally, the evaluation of the AFDS scoring models demonstrated that the greatest accuracy in predicting an FM diagnosis was provided by a combination of AFDS patient items and the AFDS clinician items that included an abbreviated (8 point) tenderpoint exam (sensitivity = 0.68, specificity = 0.82). Sensitivity of the ACR-FDC and the LFESSQ was 0.87 and 0.86, respectively, with specificity of 0.62 for the ACR-FDC and 0.49 for the LFESSQ. Most participants reported an overall preference for the AFDS (49%) over the ACR-FDC (22%) or LFESSQ (16%). Mean (SD) time to completion for each questionnaire was 5.2 minutes (3.9) for the AFDS patient items, 2.7 (2.8) for the ACR-FDC, and 1.8 (3.4) for the LFESSQ.
Conclusion: The study results indicate that the AFDS holds promise for identifying patients with FM and has measurement properties better than or similar to existing tools. The AFDS warrants further evaluation for use in the primary care setting.
Disclosure: S. Martin, Pfizer Inc, 9 ; C. Coon, Pfizer, 9 ; L. McLeod, Pfizer Inc, 3 ; A. Chandran, Pfizer Inc, 3 ; L. M. Arnold, Eli Lilly and Company, 2, Pfizer Inc, 2, Cypress Biosciences, Inc., 2, Boehringer Ingelheim, 2, Forest Laboratories, 2, Novartis Pharmaceutical Corporation, 2, Eli Lilly and Company, 5, Pfizer Inc, 5, Cypress Biosciences, Inc., 5, Forest Laboratories, 5, Takeda , 5, Astra Zeneca, 5, Sanofi-Aventis Pharmaceutical, 5, Grunenthal, 5, Johnson & Johnson, 5 .