Rasch analysis of the short form 8-item Parkinson's Disease Questionnaire (PDQ-8)

Purpose To evaluate the Italian version of the 8-item Parkinson's Disease Questionnaire (PDQ-8)-a subset of PDQ-39 (a 39-item health-related quality of life instrument for subjects with Parkinson's Disease [PD])-through classical psychometric techniques and Rasch analysis. Methods Two convenience samples (100 PD subjects each) were observed consecutively from 2004 to 2006. One group completed the PDQ-8 nested within PDQ-39, the other, the stand-alone PDQ-8. Results Once verified that the two independent samples came from the same population and showed consistent item calibrations using differential item functioning analysis, the two groups were combined. Cronbach's alpha was 0.72. According to Rasch analysis, the response scale of PDQ-8 could be simplified into a 3-category rating scale. After that, all the PDQ-8 items fitted the construct that the scale was intended to measure. Item separation reliability of PDQ-8 was 0.98 and person separation reliability was 0.70. Principal component analysis on the standardized residuals suggested a minor departure in the data from Rasch criteria (multidimensionality) and some marginal inter-item dependency. Conclusions The PDQ-8 embedded in the PDQ-39 presented psychometric properties similar to the stand-alone PDQ-8. Our results, while consistent with previous classical psychometric analyses, add information on the meaningfulness of PDQ-8 in people with PD. In particular, a simplification of its rating scale is recommended. Moreover, additional analyses should be performed in order to further check unidimensionality and local dependence, and try to improve item selection and scaling properties of the questionnaire. In order to use the PDQ-8 for clinical decision-making in reference to individuals, its reliability should first be increased.