Comparative evaluation of three semi-quantitative radiographic grading techniques for knee osteoarthritis in terms of validity and reproducibility in 1759 X-rays: report of the OARSI-OMERACT task force. Extended report

Objective: The objective of this work was to compare the measurement properties of three categorical X-ray scoring methods of knee osteoarthritis (OA), both on semiflexed and extended views. Methods: In data obtained from trials and cohorts, X-rays were graded using Kellgren and Lawrence (KL), the OA Research Society International (OARSI) joint space narrowing score, and measurement of joint space width (JSW). JSW was analyzed as a categorical variable. Construct validity was assessed through logistic regression between X-ray stages and Western Ontario and McMaster Universities OA Index. Inter-observer reliability was assessed in 50 subjects for extended views by weighted kappa. Intra-observer reliability and sensitivity to change were assessed separately for extended and semiflexed views in 50 patients who had both views performed, over a 30-month interval, by weighted kappa and standardized response mean (SRM). Results: Extended views were available from three trials and two cohorts (1759 X-rays), including one trial in which both extended and semiflexed views (antero-posterior) were obtained. Correlation with clinical parameters was low for the three scoring methods, except for the single community-based cohort. Inter-rater reliability was higher for categorical JSW in extended views (kappa, 0.86 vs 0.56 and 0.48 for KL and OARSI, respectively). Intra-rater reliability was higher for categorical JSW, both in extended views (0.83 vs 0.61 and 0.71) and in semiflexed views (0.89 vs 0.50 and 0.67). Sensitivity to change was also higher for categorical JSW, particularly in semiflexed views (SRM, 0.49 vs 0.22 and 0.34). Conclusion: These results indicate categorical JSW, in particular on semiflexed views, may be the preferred method to evaluate structural severity in knee OA clinical trials.