Summary

We created a statistical method to convert KOOS scores to IKDC-SKF scores pre- and post-ACLR using three large datasets to enable more rigorous comparisons and pooling for meta-analysis.

Abstract

The variance of patient reported outcomes measures (PROM) used in both clinical and research practice limits the comparison of outcomes and prevents pooling of data for meta-analysis. Two commonly used PROM in anterior cruciate ligament (ACL) registries and cohorts are the Knee Injury Osteoarthritis Outcomes Survey (KOOS) and International Knee Documentation Committee-Subjective Knee Form (IKDC-SKF), but few studies collect or report both scores. Our objective was to create a statistical method to convert averaged KOOS scores to IKDC-SKF scores to enable more rigorous comparisons and pooling for meta-analysis.

Methods

We used equipercentile equating methods to create a statistical crosswalk in one ACL cohort at three time-points: pre-ACL reconstruction (ACLR) and 24- and 72-months after ACLR; this was validated in two other ACL cohorts at similar time-points: pre-ACLR and 24- and 60-months post-ACLR.

Results

We observed high correlations (r=0.81-0.90), unidimensionality (first to second eigenvalues= 8.7-13.3), and subpopulation invariance (root expected mean squared difference=0.009-0.017). The smallest disagreements between crosswalked and true scores was using the 24-month scores; these had a bias of less than 0.1 standard deviation unit.

Conclusion

Our crosswalk is statistically merited and accurately converts group level average KOOS scores to IKDC-SKF scores. This tool will allow for more comparisons and meta-analyses of outcomes after ACL reconstruction, improving our treatment of and outcomes after ACL injury and reconstruction.