Article Text

Download PDFPDF
PP2 Statistical re-evaluation and modification of The Hip and Groin Outcome Score (HAGOS) questionnaire using the Rasch Model of Itemised Response Theory (IRT)
  1. Sumon S Huq1,
  2. Dylan Morrissey2
  1. 1Department of Orthopaedics, National University Hospital, Singapore
  2. 2Centre of Sport and Exercise Medicine, Queen Mary, University of London

Abstract

The Hip and Groin Outcome Score (HAGOS) is a patient-reported outcome measure (PROM) with 6 subscales assessing chronic groin pain. It was validated on a small sample of Danish subjects using Classical Test Theory. No validation has been done in English. This study addresses this using a larger sample and the Rasch model of Itemised Response Theory, an increasingly popular validation method that converts ordinal data to interval data. 190 English-speaking, active symptomatic adults completed the HAGOS for analysis of Rasch parameters of unidimensionality and fit for each subscale. Items not fitting the model were either modified by collapsing Likert scale categories or removed from its subscale which was then reanalysed. The process was repeated until the subscale showed good fit and unidimensionality. Rasch principle component analysis (PCA) was used to detect alternate dimensions within each subscale. Person separation reliability and Cronbach’s α were calculated for test-retest reliability and internal consistency respectively for all subscales. Only subscales for physical activity and activities of daily living revealed good overall fit to the Rasch model. All other subscales required modification with items removed from the symptoms, quality of life and sport and recreation subscales. PCA of the pain subscale revealed an alternative dimension worth 2.1 Eigenvalues, onto which 4 items were loaded, all related to weight bearing activities. As a result the pain subscale was further analysed as 2 separate subscales for pain on weight bearing and general pain. Person separation reliability ranged between 0.73–0.91 and Cronbach’s α between 0.81–0.91 for all modified subscales suggesting good test-retest reliability and internal consistency respectively. Initial Rasch analysis reveals that the HAGOS may not be unidimensional in its measurement properties. Further analysis suggests that, with modifications, this may be achieved although further validation is needed.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.