CRITICAL EVALUATION OF THE MIRELS RATING SYSTEM FOR IMPENDING PATHOLOGIC FEMUR FRACTURES

David Prakash, BS, Timothy Damron, MD, William Grant, PhD

PURPOSE: This project examines Mirels’ rating system of impending pathologic fracture risk for reliability/reproducibility among physicians of varying experience.

METHODS: A set of 12 clinical histories and corresponding radiographs was developed from existing patients with pathologic lesions of the femur from metastatic disease or myeloma. A questionnaire for each case was completed by 40 participants from four experience levels: orthopedic residents (10), musculoskeletal radiologists (10), orthopedic attendings (10), and fellowship trained practicing orthopedic oncologists (10). Statistical analysis including Kappa analysis was performed by a statistician.

RESULTS: Overall, total scores for the 12 cases ranged from 7 to 12 points. The score component that exhibited the greatest variability was pain (mean SD 0.77), followed by size (SD 0.73), type (SD 0.65), and site (SD 0.43) in order of diminishing variability. Kappa analysis showed very good agreement for site, good agreement for type, fair agreement for pain, and poor agreement for size. There was no significant difference in overall scores between experience levels based upon an F (variance ratio) value of 0.1199 (p = 0.95) in one way analysis of variance. Independent of Mirels definition of impending fracture, the 120 case evaluations per experience level were considered impending by residents in 106, attendings in 97, and oncologists in 85 cases. Surgical stabilization was recommended for 106 by residents, 89 by attendings, and 84 by oncologists. For the eight cases in which prophylactic stabilization was not performed prior to death (6 cases) or fracture (2 cases), the rate of correct decision making regarding prophylactic stabilization was 39% for the residents, 54% for the radiologists and for the attendings, and 57% for the oncologists.

CONCLUSIONS: Overall, the Mirels evaluation system exhibited little interobserver error within or between experience levels despite the clearly subjective nature of the pain and the size parameter ratings. However, those with increased experience were less likely to adhere strictly to Mirels definition of an impending pathologic fracture and recommended indications for prophylactic stabilization. As a result, those with greater experience had better accuracy in predicting the appropriate treatment.