RESUMEN
PURPOSE: The goal of the study was to determine the accuracy and reliability of arthroscopic percent area estimates in a plastic knee simulation model. A second goal was to determine the effect of lesion location within the knee and lesion size on accuracy and reliability. TYPE OF STUDY: Cross-sectional study of arthroscopic estimates of cartilage lesion size. METHODS: Three experienced arthroscopists performed 3 sets arthroscopic percent area estimates in 5 different plastic knees. Each knee had lesions drawn on 5 surfaces (patellar, medial and lateral femoral condyle, medial and lateral tibial plateaus). Accuracy and reliability were studied using Bland and Altman limits of agreement (LOA) and intraclass correlation coefficients. RESULTS: There was a strong tendency to overestimate lesion size by over 100% on the femoral and patellar surfaces. Intraobserver and interobserver reliabilities were generally poor. The range for the 95% LOA (+/- 1.96 standard deviation [SD] of the difference scores) between repeated measurements was almost 6 times the size of the lesion itself. Reliability of estimates was poorest for the largest lesions and worse at femoral, lateral tibial, and patellar sites. CONCLUSIONS: Assessments of arthroscopic measurements using LOA found that accuracy and reliability were generally poor, although results were better at the medial tibial plateau and for smaller lesions. In spite of these problems, arthroscopy remains a promising measurement tool because it permits physical assessment of cartilage integrity. This study sets the foundations for improvement in techniques of arthroscopic measurement of cartilage lesion size.