• Karl Bang Christensen, Kristian Thorborg, Per Hölmich, and Mikkel Bek Clausen.
• Section of Biostatistics, Department of Public Health, University of Copenhagen, Copenhagen, Denmark. kach@sund.ku.dk.
• Qual Life Res. 2019 Mar 1; 28 (3): 795-800.

PurposeThe psychometric properties of the shoulder pain and disability index (SPADI) have been extensively evaluated using classical test theory, but very few studies have applied Rasch analysis. The purpose of this study was to validate the Danish version using Rasch analysis.MethodsResponses to the SPADI from 229 patients (48% female, mean age 54.5) were included in the Rasch analysis. Overall fit, individual item fit, local response dependence, dimensionality, targeting, reliability, and differential item functioning (DIF) were examined.ResultsAfter iterative analyses, good fit to the Rasch model was observed, with acceptable targeting and uni-dimensionality. SPADI should be reported as two separate subscales: Pain and Functional Disability. The pain subscale initially demonstrated misfit due to local dependence and DIF, but a log linear Rasch model showed good fit to the Rasch model with acceptable targeting and uni-dimensionality. A six-item version of the disability subscale exhibited adequate fit in the Danish version. The same items were also found to fit the Rasch model in the English version.ConclusionsThe measurement properties of the Danish SPADI are similar to those of the English version. SPADI should be reported as two separate subscales. For the pain subscale, DIF with respect to age was disclosed, but the impact was small. The eight-item disability subscale did not fit the Rasch model. A six-item version of the disability subscale exhibited adequate fit in the Danish version. The same items were also found to fit the Rasch model in the English version.

17 keywords...

hide…