The Oswestry Disability Index, confirmatory factor analysis in a sample of 35,263 verifies a one-factor structure but practicality issues remain

Abstract

© 2017, Springer-Verlag GmbH Germany. Purpose: To analyze the factor structure of the Oswestry Disability Index (ODI) in a large symptomatic low back pain (LBP) population using exploratory (EFA) and confirmatory factor analysis (CFA). Methods: Analysis of pooled baseline ODI LBP patient data from the international Spine Tango registry of EUROSPINE, the Spine Society of Europe. The sample, with n = 35,263 (55.2% female; age 15–99, median 59 years), included 76.1% of patients with a degenerative disease, and 23.9% of the patients with various other spinal conditions. The initial EFA provided a hypothetical construct for consideration. Subsequent CFA was considered in three scenarios: the full sample and separate genders. Models were compared empirically for best fit. Results: The EFA indicated a one-factor solution accounting for 54% of the total variance. The CFA analysis based on the full sample confirmed this one-factor structure. Sub-group analyses by gender achieved good model fit for configural and partial metric invariance, but not scalar invariance. A possible two-construct model solution as outlined by previous researchers: dynamic-activities (personal care, lifting, walking, sex and social) and static-activities (pain, sleep, standing, travelling and sitting) was not preferred. Conclusions: The ODI demonstrated a one-factor structure in a large LBP sample. A potential two-factor model was considered, but not found appropriate for constructs of dynamic and static activity. The use of the single summary score for the ODI is psychometrically supported. However, practicality limitations were reported for use in the clinical and research settings. Researchers are encouraged to consider a shift towards newer, more sensitive and robustly developed instruments.