Rasch Measurement Model Supports the Unidimensionality and Internal Structure of the Arabic Oswestry Disability Index

TL;DR

The study confirms that the Arabic version of the Oswestry Disability Index is a valid tool for measuring disability in lower back pain patients.

Contribution

The paper validates the unidimensionality and internal structure of the Arabic ODI using the Rasch measurement model.

Findings

The Arabic ODI fits the Rasch model with non-significant chi-square statistics and acceptable fit residuals.

Items 2, 7, and 10 showed disordered thresholds, and item 9 showed potential bias by sex.

The t-test method confirmed the scale’s unidimensionality.

Abstract

Background/Objectives: The objective of this study was to assess the unidimensionality and internal structure of the Arabic version of the Oswestry Disability Index (ODI) in patients with lower back pain (LBP) using the Rasch measurement model. Methods: Patients with LBP (N = 113) completed the Arabic ODI during their first visit to physical therapy departments. The Arabic ODI was examined by assessing its fit to the requirements of the Rasch measurement model. Chi-square statistics for item–trait interaction alongside mean item and person fit residuals were used for overall model fit assessment. Additionally, the analysis included assessments for the fit of individual items, the sequence of thresholds, local dependency, unidimensionality using the t-test method, and differential item functioning (DIF) by sex, age, chronicity, and the presence of radiating pain. Results: The overall fit of the Arabic ODI to the Rasch measurement model was supported by non-significant Chi-square statistics (χ2 = 25.32, p = 0.19) and acceptable mean item and person fit residuals. All items showed acceptable fit (standardized fit residual −1.89 to 1.62) with no violation of local item independence. The t-test method supported the scale’s unidimensionality. The ODI showed good internal consistency with a person separation index of 0.85, with good overall targeting of item thresholds to the participants’ lower back function. Items 2, 7, and 10 showed disordered thresholds, and potential bias by sex was detected in item 9 (social life). Conclusions: The Arabic ODI is a unidimensional measure valid for assessing disability due to low back pain; however, indications of the inappropriate functioning of some response options along with potential bias by sex need to be revisited.

Figures