Construct validity of a 12-item WOMAC for assessment of femoro-acetabular impingement and osteoarthritis of the hip

Abstract

Evaluation of the internal construct validity of the Western Ontario and McMaster Universities (WOMAC) Osteoarthritis Index adapted for use in patients with femoro-acetabular impingement (FAI) and osteoarthritis (OA) of the hip. Distribution of a German version of WOMAC to patients upon first consultation. Patients with FAI [n=100, mean age 31.7 years, standard deviation (SD) 9.7] and OA (n=57, mean age 60.3 years, SD 11.7) and without comorbidities or prior hip surgery were included and compared to age- and gender-matched control population to FAI (n=200, mean age 32.6 years, SD 5.6). WOMAC data of 157 questionnaires were evaluated by Rasch analysis using RUMM2020 software. Summation of total WOMAC shows misfit to the Rasch model as well as multidimensionality. While the pain subset shows adequate fit and is unidimensional, item reduction is required to fit a unidimensional subset of functional items to the Rasch model. Summating the two fitting subsets yields again slight model misfit and multidimensionality requiring further item reduction. Finally, a 12-item version of the total WOMAC shows good model fit and unidimensionality, i.e., internal construct validity, for assessment of patients with FAI and OA without differential item functioning (DIF). A person separation index (PSI)=0.93 indicates a high internal consistency reliability for the 12-item subscale. Scores for FAI are significantly higher than control (P<0.001, effect size 0.71) and lower than OA group (P<0.001, effect size 0.45). Adequate statistical power is shown discriminating the three groups, therefore indicating some evidence also for external construct validity. The WOMAC as a total construct is multidimensional and summating the subsets into a total score is not valid. The reduced 12-item WOMAC is demonstrated to have internal construct validity for assessing patients with FAI and OA on the same scale and high internal consistency reliability. Discrimination of the groups with adequate statistical power also indicates external construct validity.