Objective: To evaluate the intra and interobserver reproducibility of 4 measuring instruments for assessing joint space width in knee osteoarthritis (OA) and to estimate the effects of patients, instrument, session order, and reader variation.
Methods: We studied 30 patients with unilateral tibiofemoral OA selected to represent a broad range of radiographic changes. Joint space width (JSW) was measured on plain anteroposterior weight bearing radiographs. Using an experimental design, 3 readers assessed JSW 3 times with 4 measuring instruments (ruler, caliper, graduated magnifying glass, digitized assessment).
Results: Intra and interobserver reproducibility was high with all measuring instruments (intraclass correlation coefficients from 0.95 to 0.98 and from 0.91 to 0.97, respectively). Analysis of variance (ANOVA) showed a patient effect (p < 10(-6)), a reader effect (p = 0.0001), an instrument effect (p = 0.0001), and a session order effect (p = 0.04). The variance component estimates were patients 55%, readers 34%, session order 2%, instruments 8%. ANOVA performed separately for each instrument showed that session order differences always represented less than 1% of the total variance. The reader component accounted for 0% of the total variance for the ruler, 2% for the digitized method, 16% for the caliper, and 18% for the graduated magnifying glass.
Conclusion: Ruler and digitized assessment have better reliability than caliper and graduated magnifying glass.