The purpose of this study was to determine the reliability and validity of the KOOS scores, using Rasch analysis for reporting outcomes following UKA.

Methods

This study was conducted at a single center with all surgeries performed by a single surgeon. Data was collected prospectively as part of quality control and retrospectively analyzed. This study was approved by an institutional review board. From 2002 to 2017, 168 medial UKA’s and 49 lateral UKA’s were performed for unicompartment osteoarthritis. Patients were included if they completed a postoperative questionnaire at a minimum of 2 years follow-up. A volunteer sampling method was used, in that all patients who presented to the tertiary orthopedic referral clinic with an unicompartment osteoarthritis were asked by a clinical staff member to participate in quality monitoring by completing a follow-up questionnaire. All subscales of the KOOS score were included in the analysis. To investigate unidimensionality, meaning that all of the items for the KOOS subscales only measure the single construct of pertaining to each specific subscale, which is an assumption of the Rasch model, a principal component analysis (PCA) was conducted, and a scree plot was produced.

Results

Person reliability was acceptable for the KOOS Sport at 0.8 and KOOS QOL at 0.85. KOOS Pain and KOOS ADL were close with 0.71 and 0.78 respectively. The item reliability was above 0.9 threshold for all subscales (KOOS Pain = 0.96; KOOS Symptom =0.93; KOOS ADL=0.96; KOOS Sport =0.91; KOOS QOL =0.97). . The person separation index was 1.56 for KOOS Pain, 0.96 for KOOS Symptoms, 1.87 for KOOS ADL, 1.98 for KOOS Sport, and 2.4 for KOOS QOL. KOOS ADL, KOOS Sport, and KOOS QOL were are near or above the recommended 2.0. Overall, KOOS Sport, KOOS ADL, KOOS QOL had good evidence of validity. To assess internal scale validity, the infit MNSQ values were calculated. KOOS Sport and KOOS QOL were the only subscales will all acceptable values. The KOOS pain subscale had poor person separation and reliability and 1 question had poor infit and outfit MNSQ. The KOOS symptom subscale also had poor person separation and reliability and 1 question had poor outfit MNSQ.

Conclusion

In this Rasch analysis of the KOOS score for use following UKA we found only 3 of the 5 subscales were sufficient to measure outcomes. The KOOS ADL subscale has acceptable person separation and reliability and 1 question had poor infit and outfit MNSQ. The KOOS Sport and KOOS QOL performed the best in the analysis with acceptable or borderline person separation and reliability. These results show that not all scales of the KOOS perform adequately when used in a population of patients following UKA.