Background/Purpose: Fibromyalgia (FM) is a chronic pain condition characterized by a constellation of symptoms and comorbidities. Consequently, FM diagnosis can be challenging and it often goes undetected. This study used electronic medical record (EMR) data to identify factors associated with FM that may facilitate earlier identification and diagnosis.

Methods: Subjects ≥18 years old who had ≥1 diagnosis code for common pain conditions were extracted from the Humedica de-identified EMR database, which has broad geographic representation and includes information on demographics, diagnoses, inpatient/outpatient encounters, medications, procedures, lab results, and select data from physicians’ notes. Records are linked using a unique patient identifier. Subjects with continuous enrollment in an integrated healthcare delivery system in 2010 and a first FM diagnosis in 2011 (cases) were compared with subjects without an FM diagnosis during 2011 (controls). Patients with an FM diagnosis prior to 2011 were excluded. FM diagnosis was based on ICD-9 code 729.1 (myalgia and myositis, unspecified). Sequential stepwise logistic regression was performed with FM diagnosis as the response variable, and demographic, clinical, and healthcare resource use as predictor variables. Variables with significant associations (P≤.05) were retained in the model and expressed as odds ratios (OR) with their 95% confidence intervals (95% CI).

Results: Subjects were 2,823 individuals with an FM diagnosis and 210,495 without an FM diagnosis in 2011. Although mean (SD) age was similar between groups, 51.4 (15.3) years for cases and 51.4 (16.4) years for controls, the FM population had more females (72.3% vs 60.4%; P<.0001), and significant differences between groups were observed for other baseline characteristics including race and healthcare resource use (higher in cases; P<.001), and the presence of specific comorbidities (higher in cases; P≤.05). The model identified 17 variables significantly associated with an FM diagnosis. The first variable (i.e., the variable with the smallest p-value when included as a predictor by itself) that made it into the model was the number of pain medication prescriptions (OR 1.03 (95% CI 1.02, 1.04), followed by the number of musculoskeletal pain conditions (OR 1.19 (95% CI 1.16, 1.23). Several clinical variables, including the presence of gastrointestinal and sleep disorders were also predictive: OR 1.38 (95% CI 1.26, 1.51) and 1.33 (95% CI 1.15, 1.55), respectively. Other healthcare resource utilization variables that entered into the model included number of outpatient visits and hospitalizations. 

Conclusion: The model identified several demographic and clinical variables as significant predictors of an FM diagnosis. These results suggest analysis of EMR data can help identify variables associated with FM in a real world setting, and may inform earlier identification of FM patients.

---

« Back to 2013 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/identifying-predictors-of-a-fibromyalgia-diagnosis-a-retrospective-electronic-medical-record-analysis/