A M Davis1, E M Badley, D E Beaton, J Kopec, J G Wright, N L Young, J I Williams. 1. Department of Physical Therapy, Graduate Department of Rehabilitation Science and Health Policy, Management and Evaluation, Toronto Rehabilitation Institute and the University of Toronto, Toronto, ON, Canada. davis.aileen@torontorehab.on.ca
Abstract
BACKGROUND AND OBJECTIVE: This study is based on secondary analysis of Western Ontario McMaster Osteoarthritis Index (WOMAC) data from a community sample over 55 years and total hip or knee arthroplasty samples presurgery and 1-year postoperative. METHODS: The WOMAC data were evaluated by Rasch analysis. Data were considered to fit the Rasch mathematical model for the pain and physical dimensions of the WOMAC if unidimensionality was confirmed by principle component analysis of the subscale and the residuals from the Rasch analysis, infit and outfit statistics were in the range of 0.80 to 1.20; if there was no differential item functioning based on gender or hip vs. knee subjects; and, if there was stability of the item logits across the three data samples. RESULTS: A three-item pain dimension (excluding night pain and pain on standing) and a 14-item physical dimension (excluding heavy domestic duties, getting in and out of the bath and getting on and off the toilet) fit the Rasch model based on these criteria. CONCLUSION: In evaluating existing health status questionnaires using Rasch methodology, it is important to evaluate relevant patient samples and longitudinal data when the measure is intended to evaluate change in status. By these criteria, a modified WOMAC questionnaire fits the Rasch model and has interval-level scaling properties.
BACKGROUND AND OBJECTIVE: This study is based on secondary analysis of Western Ontario McMaster Osteoarthritis Index (WOMAC) data from a community sample over 55 years and total hip or knee arthroplasty samples presurgery and 1-year postoperative. METHODS: The WOMAC data were evaluated by Rasch analysis. Data were considered to fit the Rasch mathematical model for the pain and physical dimensions of the WOMAC if unidimensionality was confirmed by principle component analysis of the subscale and the residuals from the Rasch analysis, infit and outfit statistics were in the range of 0.80 to 1.20; if there was no differential item functioning based on gender or hip vs. knee subjects; and, if there was stability of the item logits across the three data samples. RESULTS: A three-item pain dimension (excluding night pain and pain on standing) and a 14-item physical dimension (excluding heavy domestic duties, getting in and out of the bath and getting on and off the toilet) fit the Rasch model based on these criteria. CONCLUSION: In evaluating existing health status questionnaires using Rasch methodology, it is important to evaluate relevant patient samples and longitudinal data when the measure is intended to evaluate change in status. By these criteria, a modified WOMAC questionnaire fits the Rasch model and has interval-level scaling properties.
Authors: M Orive; A Anton-Ladislao; S García-Gutiérrez; C Las Hayas; N González; J Zabala; J M Quintana Journal: Osteoporos Int Date: 2015-08-05 Impact factor: 4.507
Authors: Natalie J Collins; Devyani Misra; David T Felson; Kay M Crossley; Ewa M Roos Journal: Arthritis Care Res (Hoboken) Date: 2011-11 Impact factor: 4.794
Authors: Amaia Bilbao; José M Quintana; Antonio Escobar; Carlota Las Hayas; Miren Orive Journal: Health Qual Life Outcomes Date: 2011-09-21 Impact factor: 3.186
Authors: Julie F Pallant; Anne-Maree Keenan; Roseanne Misajon; Philip G Conaghan; Alan Tennant Journal: Health Qual Life Outcomes Date: 2009-04-29 Impact factor: 3.186
Authors: Liseth Siemons; Peter M Ten Klooster; Erik Taal; Cees Aw Glas; Mart Afj Van de Laar Journal: BMC Musculoskelet Disord Date: 2012-10-31 Impact factor: 2.362