Rationale and objectives: To assess the intraobserver reliability of three methods used frequently to evaluate joint destruction in rheumatoid arthritis: the Sharp method, the Larsen method, and the carpo:metacarpal ratio.
Methods: One observer analyzed twice within a 6-week interval 71 radiographs from patients with rheumatoid arthritis. Reliability was estimated by intraclass correlation coefficient (R) and by Altman-Bland graphical method. Correlations were examined by the Spearman's coefficient (r).
Results: The intraobserver reliability of each method appeared satisfactory with a good result for the Sharp method (R = 0.97). The correlation was strong (r > 0.80) between the results obtained by Sharp's and Larsen's methods and weaker between the results of the two former methods and the carpo: metacarpal ratio.
Conclusions: Sharp's method should be used preferentially in studies evaluating the radiologic changes in rheumatoid arthritis over time, especially in clinical trials. The carpo:metacarpal ratio may be considered as a complementary method, when wrist destruction is of conceptual importance.