Extension from inpatients to outpatients: validity and reliability of the Oxford Knee Score in measuring health outcomes in patients with knee osteoarthritis

Abstract

To validate the Oxford Knee Score (OKS) in outpatients with knee osteoarthritis (OA). Eligible patients were interviewed using a pretested questionnaire containing the OKS, the Short Form (SF)-6D, and the EuroQol Group 5-Dimension Self-Report Questionnaire score (EQ-5D). Reliability was assessed using Cronbach's alpha, dimensionality using principal component factor analysis and item-total correlations, convergent and discriminant construct validity using expected correlations between the OKS and the SF-6D and the EQ-5D. The OKS were well accepted by patients in a pilot testing. When administered to a convenient sample of 187 patients with knee OA (mean age 64 years, 74% female, mean duration of OA 7.6 years), Cronbach's alpha exceeded 0.8 and factor analysis yielded two factors with eigenvalues > 1. Hypothesized item-total correlations (ρ ≥ 0.4) were observed for all items. Convergent construct validity was supported by the presence of hypothesized moderate to strong correlations between the OKS and SF-6D physical functioning, role limitation, social functioning and pain, EQ-5D mobility, self-care, usual activities, and pain/discomfort, and mobility visual analogue scale (ρ = 0.47-0.82). Discriminant construct validity was not satisfactory. The OKS correlated weakly with SF-6D vitality (ρ = 0.35), but strongly with SF-6D mental health and EQ-5D anxiety/depression (ρ = 0.51 and 0.41, respectively). The OKS has demonstrated good psychometric properties and thus can be considered a reliable and valid measurement for outpatients with OA.