Related Experiment Videos
Crosswalk and correlation between the Oxford Knee Score (OKS), Knee Society Score (KSS) and Knee Injury and
Shaun Kai Kiat Chua1, Linus Ren Hao Tan2, Chien Joo Lim3
1Department of Orthopaedic Surgery, Woodlands Hospital, National Healthcare Group, 17 Woodlands Dr 17, Singapore, 737628, Singapore. schua041@e.ntu.edu.sg.
Background:
Significant heterogeneity exists in the patient-reported outcome measures (PROMs) used to assess outcomes in knee osteoarthritis (KOA), with limited evidence exploring the correlation and conversion among the various PROMs. Our study aims to explore the correlation and conversion between the Oxford Knee Score (OKS), Knee Injury and Osteoarthritis Outcome Score (KOOS-12) and Knee Society Score (KSS) in patients with non-operatively treated KOA.
Methodology:
This was a prospective cohort study of 131 patients who received non-operative treatment (standardized physical therapy and analgesia) for KOA between July 2020 and January 2024. Three validated PROMs for KOA (OKS, KOOS-12 and KSS) were measured for patients at baseline, 3-months, 6-months, 1-year and 2-years follow-up. Overall, 498 complete paired outcome scores (KSS/OKS/KOOS-12 at a single time point) were included for analysis. The data were then randomly split into a derivation (75%) and internal validation (25%) cohort. The correlation between the outcome scores were calculated, and linear regression model was used to obtain conversion equations between the measured outcomes. Internal validation of the derived conversion equation was performed subsequently by comparing the mean difference between the equation-derived and actual scores.
Results:
A significant strong positive correlation was found between KOOS-12 and OKS (coefficient:0.79;p < 0.001), and a significant moderate positive correlation was found between the KOOS-12 and KSS (coefficient:0.50;p < 0.001). A significant moderate positive correlation was found between the OKS and KSS (coefficient:0.67;p < 0.001). Six conversion equations derived for the measured PROMs had low mean difference between the equation-derived and actual scores (0.31-1.68). Moderate to strong correlation was found between the actual and equation-derived scores, with weakest strength observed in the conversion between KOOS-12 and KSS (coefficient:0.53) and the strongest strength observed in the conversion between KOOS-12 and OKS (coefficient:0.80).
Conclusions:
On internal validation, we found good correlation between the mean actual and equation-derived outcome scores despite significant within-individual variability. The conversion equations derived in this study can facilitate pooling and comparison of studies using heterogenous outcomes measures and meta-analysis to increase statistical power and address novel questions in KOA. Nonetheless, the equations should not be utilized to predict individual scores from one to another due to high within-individual variability.
Clinical Trial Number:
Not applicable.
