Reliability and validity of Oswestry Disability Index among patients undergoing lumbar spinal surgery

Abstract

<p><strong>Background: </strong>The objective of this study was to explore the internal consistency and factor structure of the Oswestry Disability Index among patients undergoing spinal surgery. The sample consisted of 1,515 patients who underwent lumbar spinal surgery at a university hospital between 2018 and 2021.</p><p><strong>Methods: </strong>The patients responded to the Oswestry Disability Index within 2 months before surgery. Cronbach's alpha was used to assess the internal consistency. The factor structure was evaluated using exploratory and confirmatory factor analyses.</p><p><strong>Results: </strong>The average age of 1,515 patients was 58.5 (SD 15.8) years and 53% were women. The mean ODI score was 43.4% (SD 17.4%). Of the patients, 68% underwent microsurgical excision of the lumbar intervertebral disc displacement or decompression of the lumbar nerve roots. The internal consistency of the Oswestry Disability Index was found to be good, with an alpha of 0.87 (95% CL 0.86 to 0.88). Exploratory factor analysis resulted in unidimensional structure. Item loadings on this retained factor were moderate to substantial for all 10 items. One-factor confirmatory factor analysis model demonstrated an acceptable fit. The correlations between the main factor "disability" and the individual items varied from moderate (0.44) to substantial (0.76). The highest correlations were observed for items "traveling", "personal care", and "social life". The lowest correlations were observed for the item "standing".</p><p><strong>Conclusions: </strong>The Oswestry Disability Index is a unidimensional and internally consistent scale that can be used to assess the severity of disability in patients undergoing lumbar spinal surgery. In the studied population, "traveling," "social life," "sex life" and "personal care" were the most important items to define the severity of disability, while "walking" and "standing" were the least important items. The generalizability of the results might be affected by the heterogeneity and modest size of the studied cohort.</p>