PRM152 - SAT-Q questionnaire: Application of rasch measurement for selecting items and improving the assessment of patient satisfaction

Abstract

In the last years, the assessment of patient satisfaction has become a field of interest for public authorities, researchers, and clinicians due to its implications specially, in the management of chronic diseases. To this purpose the SAT-Q is a questionnaire for measuring patient satisfaction which has been developed and validated from a Classical Test Theory (CTT) approach, resulting in a 18 item version. The objective of this research focuses on the application of the Rasch model in selecting and testing the most appropriate items for the assessment of the satisfaction related to pharmacological treatments. The sample consisted of a total of 202 patients diagnosed with chronic illness who were following a pharmacological regimen for at least one year. Data were analyzed using the Rasch Partial Credit Model. Parameter estimation, model-fit data and separation statistics were computed and compared to CTT validation, in order to select those items with better psychometrical properties. Misfiting items were deleted, resulting in a 10 items version. This short form of the SAT-Q showed good model fit and represented an accurate measure of the construct. Infit MNSQ ranged from 0.75 to 1.51 (SD=0.25) and outfit MNSQ ranged from 0.68 to 2.24 (SD=0.45). Person separation index and item reliability were 1.90 and 0.97 respectively. Principal Component Analysis (PCA) also distinguished a main component that accounted for 54.5% of total variance. Score correlation with CTT validation version was 0.83. On the other hand, item-person map showed that the test was off-target and didn’t cover the wide range of the ability scale. SAT-Q short form version has adequate psychometric properties in terms of goodness-of-fit-test, reliability and validity. Future research is recommended to improve discrimination between different degrees of satisfaction and reducing the ceiling effect associated with the satisfaction measures.