The Romanian translation of the KOOSJR proved valid, reliable, consistent and reproducible in patients with end stage OA undergoing total knee replacement. The Cronbach’s alpha and ICC were comparable to recently published literature regarding KOOS translations: Spanish Cronbach's 0.78-0.93 and ICC 0.76 to 0.91; Finnish Cronbach’s 0.79-0.96 and ICC 0.73-0.86; Chinese Cronbach’s 0.76-0.97 and ICC 0.89-0.95 and Greek ICC 0.76-.89 [12-15].
A direct comparison to similar translations of KOOSJR is difficult since data is only available for the full KOOS. This later has 7 questions for symptoms, 9 for pain, 17 for activities of daily living, 5 for sports and 4 for quality of life for a total of 42 5 point Likert scale items. It is freely available, self-explanatory, comprehensive, widely used for knee injuries leading to arthritis or OA and also includes the proprietary WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) [5,9].
Nevertheless, for routine clinical use especially in elderly, shorter scores are favored for compliance and efficiency. The 7-question standardized KOOSJR was proven to be just as reliable in patients undergoing total knee replacement. Other shortened versions of the KOOS have been proposed: the 7 question KOOS-PS (physical activity) and recently the 12 question KOOS-12 . A single question – the M-SANE asked patients to rate their native or prosthetic knee on a scale from 0 to 10. It corelated strong to moderate to KOOSJR and PROMIS (Patient-Reported Outcomes Measurement Information System) physical component .
The PROMIS uses computer adaptive testing, where algorithms select the best questions from a larger database. It is aimed at offering a unified tool for use in different pathologies as well as integrate disease specific points to activities of daily living. Its responsiveness is comparable to KOOSJR and HOOSJR (Hip disability and Osteoarthritis Outcome Score for Joint Replacement) in patients undergoing total joint arthroplasty .
Our study has several limitations. Firstly, we did not use the entire KOOS and secondarily subtract the KOOSJR. We felt that the full score might have been rather cumbersome to use in current elderly population undergoing knee arthroplasty in Romania and the simplified KOOSJR was proven to offer comparable usefulness [3-5,16]. Furthermore, we did not test responsiveness, by including a timepoint test after surgery and the translated Romanian IKDC form, the strongest comparator for validity is currently undergoing validation. The original KSS score had several limitations including high variability, acknowledged by the developer and addressed by complete revision in 2011. When our study was designed, the original KSS had been the standard of use in our clinic and the new score was not yet available free of charge. This may justify the very weak correlation found in our study between the KSS patient form and function and all other tested scores.
Knee OA is the common endpoint for a multitude of pathologies. Until present, there are no disease modifying drugs and treatment of early stages is mainly symptomatic. Total knee arthroplasty has become the mainstay for advanced disease in the elderly for many years, yet still some patients exhibit unreliable improvements. Prediction models may be one way to stratify patients at risk of poor outcomes. A group of researchers found low Oxford knee scores, poverty, increased body mass index, anxiety and depression to predict worse outcomes. In addition, there are also local factors such as impaired physical status and previous knee arthroscopy that are negative predictors of outcome. Contrarily, a fixed flexion deformity and absence of the anterior cruciate ligament were associated with postoperative improvement . Machine learning algorithms are still at the beginning but show promising ability to predict which patients will achieve increased improvement after knee replacement .