Abstract
The aim of this study was to verify the single-factor structure of the joint replacement
version of the Knee Injury and Osteoarthritis Outcome Score (KOOS-JR) and examine
its measurement properties in the context of Rasch analysis in patients with end-stage
osteoarthritis of the knee (KOA) awaiting total knee replacement (TKR). The study
design was retrieval of prospectively collected clinical data. The data were extracted
from the presurgery visit for individuals with KOA who were scheduled for primary
TKR at a tertiary care hospital. Those who were scheduled for revision of TKR had
any other lower extremity injury or surgery during 6 months prior to the presurgery
visit, or those who had reported pre-existing neurological impairments affecting the
lower extremity functions were excluded during data extraction. The assumptions of
Rasch analysis that were examined included the test of fit, fit of residuals, ordering
of item thresholds, Pearson separation index, differential item functioning (DIF),
dependency, and unidimensionality. The main outcome measure was KOOS-JR. Data were
extracted for 283 patients, including 112 men and 160 women, from clinical charts.
The KOOS-JR demonstrated good overall fit to the Rasch model. However, it failed to
meet the assumption of unidimensionality. None of the items demonstrated DIF or concerns
with response thresholds. Person–item threshold distribution indicated that the score
for KOOS-JR overestimated person traits with floor and ceiling effects. Reliability
statistics were equal to 0.9, suggesting that seven items within the KOOS-JR were
internally consistent and reliable. The hypothetical unidimensional KOOS-JR could
not be reproduced in our sample in that KOOS-JR had a latent construct. Future research
should perform exploratory factor analysis to examine this latent construct.
Keywords
knee replacement - KOOS - Rasch analysis - knee pain
