Using Rasch Analysis to Evaluate the Reliability and Validity of the Swallowing Quality of Life Questionnaire: An Item Response Theory Approach


Cordier R., Speyer R., Schindler A., Michou E., Heijnen B. J. , Baijens L., ...Daha Fazla

DYSPHAGIA, cilt.33, ss.441-456, 2018 (SCI İndekslerine Giren Dergi) identifier identifier identifier

  • Cilt numarası: 33 Konu: 4
  • Basım Tarihi: 2018
  • Doi Numarası: 10.1007/s00455-017-9873-4
  • Dergi Adı: DYSPHAGIA
  • Sayfa Sayıları: ss.441-456

Özet

The Swallowing Quality of Life questionnaire (SWAL-QOL) is widely used clinically and in research to evaluate quality of life related to swallowing difficulties. It has been described as a valid and reliable tool, but was developed and tested using classic test theory. This study describes the reliability and validity of the SWAL-QOL using item response theory (IRT; Rasch analysis). SWAL-QOL data were gathered from 507 participants at risk of oropharyngeal dysphagia (OD) across four European countries. OD was confirmed in 75.7% of participants via videofluoroscopy and/or fiberoptic endoscopic evaluation, or a clinical diagnosis based on meeting selected criteria. Patients with esophageal dysphagia were excluded. Data were analysed using Rasch analysis. Item and person reliability was good for all the items combined. However, person reliability was poor for 8 subscales and item reliability was poor for one subscale. Eight subscales exhibited poor person separation and two exhibited poor item separation. Overall item and person fit statistics were acceptable. However, at an individual item fit level results indicated unpredictable item responses for 28 items, and item redundancy for 10 items. The item-person dimensionality map confirmed these findings. Results from the overall Rasch model fit and Principal Component Analysis were suggestive of a second dimension. For all the items combined, none of the item categories were 'category', 'threshold' or 'step' disordered; however, all subscales demonstrated category disordered functioning. Findings suggest an urgent need to further investigate the underlying structure of the SWAL-QOL and its psychometric characteristics using IRT.