Item Response Theory for Psychometric Properties of the SNOT-22 (22-Item Sinonasal Outcome Test)

Objective The SNOT-22 (22-item Sinonasal Outcome Test) is a high-quality outcome measure that assesses chronic rhinosinusitis–specific quality of life. The aim of this study was to gain greater insight into the information provided by the SNOT-22 by determining its item-based psychometric properties. Study Design Retrospective cohort study. Setting Tertiary care academic centers. Methods This study used a previously described data set of the SNOT-22 completed by 800 patients with chronic rhinosinusitis. Item response theory graded response models were used to determine parameters reflecting item discrimination, difficulty, and information provided by each item toward the SNOT-22 subdomain to which it belonged. Results The unconstrained graded response model fitted the SNOT-22 data best. Item discrimination parameters and total information provided showed the greatest variability within the nasal subdomain, and the item related to sense of smell/taste demonstrated the lowest discrimination and provided the least amount of information overall. The dizziness item provided disparately lower total information and discrimination in the otologic/facial pain subdomain. Items in the sleep and emotional subdomains generally provided high discrimination. While items in the nasal, sleep, and otologic/facial pain subdomains spanned all levels of difficulty, emotional subdomain items covered higher levels of difficulty, indicating greater information provided at higher levels of disease severity. Conclusion The item-specific psychometric properties of the SNOT-22 support it as a high-quality instrument. Our results suggest the need and possibility for revision of the smell/taste dysfunction item, for example its wording, to improve its ability to discriminate among the different levels of disease burden.