Confirmatory Factor Analysis and Multiple Linear Regression of the Neck Disability Index: Assessment If Subscales Are...

Because of previously published recommendations to modify the Neck Disability Index (NDI), we evaluated the responsiveness and dimensionality of the NDI within a population of adult whiplash-injured subjects. The purpose of the present study was to evaluate the responsiveness and dimensionality of the NDI within a population of adult whiplash-injured subjects.

Subjects who had sustained whiplash injuries of grade 2 or higher completed an NDI questionnaire. There were 123 subjects (55% female, of which 36% had recovered and 64% had chronic symptoms. NDI subscales were analyzed using confirmatory factor analysis, considering only the subscales and, secondly, using sex as an 11th variable. The subscales were also tested with multiple linear regression modeling using the total score as a target variable.

When considering only the 10 NDI subscales, only a single factor emerged, with an eigenvalue of 5.4, explaining 53.7% of the total variance. Strong correlation (> .55) (P < .0001) between all variables was found. Multiple linear regression modeling revealed high internal consistency with all coefficients reaching significance (P < .0001). The 4 NDI subscales exerting the greatest effect were, in decreasing order, Sleeping, Lifting, Headaches, and Pain Intensity.

A 2-factor model of the NDI is not justified based on our results, and in this population of whiplash subjects, the NDI was unidimensional, demonstrating high internal consistency and supporting the original validation study of Vernon and Mior.