Is the Interpretation of Radiographic Knee Arthritis Consistent Between Orthopaedic Surgeons and Radiologists?

Knee radiographs are often examined independently by both radiologists and orthopaedic surgeons when evaluating osteoarthritis (OA). While multiple systems have been described, formal classification systems are infrequently used in clinical practice and documentation. Instead, providers commonly describe knee OA on radiographs as “mild,” “moderate,” or “severe,” with loose and unclear interpretations. From a patient’s perspective, inconsistent reading and charting of knee OA severity can have financial and psychological implications, such as prior authorization denial, as well as anxiety-provoking uncertainty with their diagnosis. The purpose of this study was to investigate the agreement between orthopaedic surgeons, musculoskeletal radiologists, and general radiologists on the severity and location of knee OA. 105 deidentified radiographs of patients presenting with knee pain were obtained. Anteroposterior (AP) and lateral radiographs were reviewed independently by two high-volume arthroplasty surgeons, two musculoskeletal radiologists, and two general radiologists. Each radiograph was classified as mild, moderate, or severe OA, mirroring the language used in the providers’ documentation. Providers were also asked to comment on the location of OA, described as medial, lateral, patellofemoral, or any combination. Agreement was calculated using Fleiss’ kappa in which values less than 0.3 were considered no true agreement, 0.3 and 0.5 weak agreement, 0.5 and 0.8 moderate agreement, and greater than 0.8 strong agreement. There was inconsistent agreement for severity and location among physicians of the same specialty and between specialties. There was moderate agreement (k = 0.513) in the assessment of patellofemoral arthritis among radiologists. Orthopaedic surgeons (k = 0.503) and musculoskeletal radiologists (k = 0.568) demonstrated moderate agreement in the perceived need for TKA, and there was moderate agreement between the two specialties (k = 0.556). All other comparisons indicate weak or no agreement. A high degree of inconsistency was found in the subjective interpretation of radiographic knee OA. Although grading systems exist, providers often document knee OA based on the terms “mild,” “moderate,” and “severe,” which was shown to have poor reliability. Utilization and adherence to an existing standardized system of interpreting knee x-rays, which can be efficiently integrated into clinical practice, is necessary to improve communication for providers, patients, and insurers.