Validation of the NHANES ADL scale in a sample of patients with report of cervical pain: factor analysis, item response theory analysis, and line item validity.

Journal Article

BACKGROUND: Few functional outcomes scales have used Item Response Theory (IRT) for validation. IRT allows individual line item validations and offers substantial advantages over classic methods of scale validation or the simplest from of IRT known as Rasch. Rasch analysis reduces data to dichotomous variables thus decreasing the sensitivity of Likert-type data responses. PURPOSE: The purpose of this study was to create an outcome scale associated with the latent trait of functioning and disability, validated using IRT, in a population with report of cervical pain. METHODS: Using the NHANES database, a recently created scale (NHANES ADL scale) was analysed using factor analysis, internal analyses of consistency, IRT, and comparison with internal measures of functioning and disability. RESULTS: The newly created NHANES ADL scale demonstrated uni-dimensionality, was internally reliable, and was correlated to internal measures of functioning and disability. Additionally, the majority of the scale items demonstrate strong discrimination and suitable thresholds. DISCUSSION: The NHANES ADL scale effectively measures physical, social, and emotional disability in patients with a cervical impairment, and may be an efficient measure of perceived limitations from working and generalized daily physical activity. CONCLUSION: The newly created NHANES ADL scale demonstrates internal consistency, unidimensionality, and line item validity. The NHANES ADL scale appears to be a useful instrument in measurement of functioning and disability in patients with report of cervical pain.

Full Text

Duke Authors

Cited Authors

  • Cook, CE; Richardson, JK; Pietrobon, R; Braga, L; Silva, HM; Turner, D

Published Date

  • August 15, 2006

Published In

Volume / Issue

  • 28 / 15

Start / End Page

  • 929 - 935

PubMed ID

  • 16861200

International Standard Serial Number (ISSN)

  • 0963-8288

Digital Object Identifier (DOI)

  • 10.1080/09638280500404263

Language

  • eng

Conference Location

  • England