Development and Validation of a Simplified Three‐Item Score for Predicting Eosinophilic Chronic Rhinosinusitis
Kao‐Tsung Lin, Yi‐Chen Tsai, Te‐Huei Yeh, Chih‐Feng Lin, Yi‐Tsen LinABSTRACT
Objective
Accurate preoperative identification of eosinophilic chronic rhinosinusitis (ECRS) is essential for treatment planning and selection of biologic therapies, particularly in Asian populations where inflammatory endotypes are more heterogeneous. We aimed to develop and validate a simplified clinical scoring system using readily obtainable parameters for predicting ECRS.
Methods
A total of 263 adults with chronic rhinosinusitis undergoing endoscopic sinus surgery were enrolled and divided into a training cohort ( n = 158) and an independent validation cohort ( n = 105). ECRS was defined histologically as ≥ 10 eosinophils per high‐power field. Candidate predictors were evaluated using receiver operating characteristic analysis and multivariable logistic regression to construct a simplified 3‐item ECRS prediction score. Diagnostic performance was compared with the JESREC score using DeLong's test.
Results
Three independent predictors were identified: blood eosinophil count ≥ 200/μL (three points), computed tomography ethmoid–maxillary (E/M) ratio ≥ 1 (2 points), and presence of nasal polyps (one point). A total score ≥ 4 predicted ECRS. In the validation cohort, the 3‐item score demonstrated good discrimination (AUC = 0.781), with sensitivity 77.3% and specificity 73.8%. Diagnostic performance was comparable to the JESREC score (AUC = 0.840; sensitivity 77.3%; specificity 72.1%), with no significant difference between models (Holm‐adjusted p = 0.206).
Conclusion
The simplified 3‐item ECRS prediction score provides a practical tool for preoperative prediction of ECRS using routinely available clinical parameters. Despite requiring fewer variables, its diagnostic performance was comparable to the JESREC score, supporting its clinical utility for rapid endotype identification and improved treatment stratification in routine practice.