Critical evaluation of Mirels' rating system for impending pathologic fractures

Abstract

This project examined the hypothesis that Mirels' rating system for impending pathologic fractures is reproducible, valid, and applicable across various experience levels and training backgrounds. Twelve true clinical histories and corresponding radiographs for patients with femoral metastatic lesions were reviewed by 53 participants from five experience levels: orthopaedic residents, musculoskeletal radiologists, orthopaedic attendings, fellowship-trained practicing orthopaedic oncologists, and radiation or medical oncologists. Each examiner provided individual and total Mirels' scores and independent determination of impending fracture using clinical judgment. A subset of seven histories without prophylactic fixation provided a natural history group. There was highly significant agreement across experience categories for overall Kappa and for the concordance for individual and overall scores. Kappa analysis showed good agreement for site, moderate agreement for type, and fair agreement for size and pain. There was no significant difference in overall scores across experience levels. The pooled odds ratio favored Mirels rating system over clinical judgment regardless of experience level. Overall sensitivity was 91% and specificity was 35%. Mirels' system seems to be reproducible, valid, and more sensitive than clinical judgment across experience levels. However, although the system is a valuable screening tool, more specific parameters are needed.