Rasch analysis of the Forgotten Joint Score in patients undergoing knee arthroplasty.

The purpose of this study was to test the hypothesis that the "Forgotten Joint Score" (FJS-12) is a unidimensional interval-level scale. Unidimensionality refers to measuring a single attribute, i.e., the single ability to forget the arthroplasty. If this property is not verified, the interpretation of the score can be confusing. Unidimensionality is an essential prerequisite of construct validity and required if FJS-12 response data are to be validly summated into a single score. Interval-level dimension is an essential prerequisite of the parametric statistics. Rasch analysis was used to test our study hypothesis. The FJS-12 questionnaire was validated in 248 unilateral knee arthroplasty patients. Successive analyses were used to select items with good psychometric qualities to constitute the new "FJS". The external validity was assessed with the KUJALA questionnaire. Quantity of relevant items was greater than 50%. Of the 12 original items, nine showed disturbed thresholds, indicating that patients were unable to discriminate among the five levels for these items. The data set was reanalyzed using a four-level scale. The new analysis indicated that the internal consistency was good (r = 0.84). Three items did not fit with the model and they were removed. The nine items of the final scale defined a unidimensional and linear measure of the forgotten joint, and showed a continuous progression in their difficulty. The perception of difficulty was group-independent. The correlation coefficient was moderate between FJS and KUJALA score (r = 0.4). This new and items reduced FJS can be used in clinical practice with good psychometric qualities. It provides a reliable tool to follow up patient's evolution and document changes related to knee arthroplasty. This valid FJS is needed in evaluating patients' assessment, one indicator of quality of care. III-Therapeutic.