Development of a logically devised line drawing atlas for grading of knee osteoarthritis.

OBJECTIVES To (a) develop an atlas of line drawings for the assessment and grading of narrowing and osteophyte (that is, changes of osteoarthritis) on knee radiographs, and (b) compare the performance of this atlas with that of the standard Osteoarthritis Research Society (OARS) photographic atlas of radiographs. METHODS Normal joint space widths (grade 0) for the medial and lateral tibiofemoral and medial and lateral patellofemoral compartments were obtained from a previous community study. Grades 1-3 narrowing in each compartment was calculated separately for men and women, grade 3 being bone on bone, grades 1 and 2 being two thirds and one third the value of grade 0. Maximum osteophyte size (grade 3) for each of eight sites was determined from 715 bilateral knee x ray films obtained in a knee osteoarthritis (OA) hospital clinic; grades 1-2 were calculated as two thirds and one third reductions in the area of grade 3. Drawings for narrowing and osteophyte were presented separately. 50 sets of bilateral knee x ray radiographs (standing, extended anteroposterior; flexed skyline) showing a spectrum of OA grades were scored by three observers, twice using the OARS atlas and twice using the drawn atlas. RESULTS Intraobserver and interobserver reproducibility was similar and generally good with both atlases, though varied according to site. All three observers preferred the line drawing atlas for ease and convenience of use. Higher scores for patellofemoral narrowing and lower scores for osteophyte, especially medial femoral osteophyte, were seen using the line drawing atlas, showing that the two atlases are not equivalent instruments. CONCLUSION A logically derived line drawing atlas for grading of narrowing and osteophyte at the knee has been produced. The atlas showed comparable reproducibility with the OARS atlas, but was discordant in several aspects of grading. Such a system has several theoretical and practical advantages and should be considered for use in knee OA studies.