Objective: Test if differential item functioning (DIF) due to gender, age, race/ethnicity or education level is present in UDysRS items.   

Background: Testing a rating scale for DIF is a core step in comprehensive validation methodology. DIF occurs for the UDysRS when the probability of item scores differs among people with similar levels of dyskinesia but belong to different groups on a secondary trait (gender, age, race/ethnicity, education).  If DIF is present, interpretation of an item score needs to include consideration of the secondary trait as well as dyskinesia severity. There are 2 types of DIF: uniform (U-DIF), where the influence on item scores by the secondary trait is constant over all levels of the main trait; and non-uniform (NU-DIF), where the influence on item scores by the secondary trait varies across levels of dyskinesia. 

Methods: Using the cross-sectional UDysRS translation database (N=3,075), we first confirmed unidimensionality of the UDysRS using CFA techniques.  We then tested the impact of gender, age (28-51, 52-75, 75-97 yrs), race/ethnicity (white vs other, Asian vs other, Hispanic vs other), and education (< 7, 7-12, >12yrs) on U-DIF and NU-DIF. We required that two independent methods, MIMIC and lordif, both identified item-specific DIF to qualify for consideration. Because very few of those patients studied had scores of 4, we collapsed scores of 3 and 4 into one category to allow the methods to converge mathematically. The DIF impact was determined by McFadden pseudo R² cut-offs (large, moderate, negligible) and considered items pertinent if they exceeded the criteria beyond negligible R² > 0.035.    

Results: No age-, race/ethnicity or education NU-DIF was identified.  Gender NU-DIF was found for historical ratings of Speech and examination for communication, both with negligible impact.  U-DIF was observed for multiple UDysRS items for gender, age, race and education, but none exceeded negligible impact.

Conclusions: Gender, age, race/ethnicity and education have no pertinent DIF impact for the UDysRS, allowing the scale to be utilized as a core outcome measure across populations with varying gender, age, racial or educational distributions. This finding strongly argues that the UDysRS is effectively capturing dyskinesia and does not function differently across groups defined by gender, age, race/ethnicity or education.

References: Goetz CG, Nutt JG, Stebbins GT. The unified dyskinesia rating scale: presentation and clinimetric profile. Movement Disorders. 2008 Dec 15;23(16):2398-403.

Goetz CG, Liu Y, Stebbins GT, Wang L, Tilley BC, Teresi JA, Merkitch D, Luo S. Gender‐, age‐, and race/ethnicity‐based differential item functioning analysis of the movement disorder society–sponsored revision of the Unified Parkinson’s disease rating scale. Movement Disorders. 2016 Dec 1;31(12):1865-73.

« Back to 2017 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/differential-item-functioning-in-the-unified-dyskinesia-rating-scale-udysrs/