The 39 item Parkinson's disease questionnaire (PDQ-39) is the most widely used patient reported rating scale in Parkinson's disease. However, several fundamental measurement assumptions necessary for confident use and interpretation of the eight PDQ-39 scales have not been fully addressed.
Postal survey PDQ-39 data from 202 people with Parkinson's disease (54% men; mean age 70 years) were analysed regarding psychometric properties using traditional and Rasch measurement methods.
Data quality was good (mean missing item responses, 2%) and there was general support for the legitimacy of summing items within scales without weighting or standardisation. Score reliabilities were adequate (Cronbach's alpha 0.72-0.95; test-retest 0.76-0.93). The validity of the current grouping of items into scales was not supported by scaling success rates (mean 56.2%), or factor and Rasch analyses. All scales represented more health problems than that experienced by the sample (mean floor effect 15%) and showed compromised score precision towards the less severe end.
Our results provide general support for the acceptability and reliability of the PDQ-39. However, they also demonstrate limitations that have implications for the use of the PDQ-39 in clinical research. The grouping of items into scales appears overly complex and the meaning of scale scores is unclear, which hampers their interpretation. Suboptimal targeting limits measurement precision and, therefore, probably also responsiveness. These observations have implications for the role of the PDQ-39 in clinical trials and evidence based medicine. PDQ-39 derived endpoints should be interpreted and selected cautiously, particularly regarding small but clinically important effects among people with less severe problems.