Introduction
Chronic rhinosinusitis (CRS) is a symptomatic inflammatory disease of the nasal and paranasal mucosalasting more than 12 weeks1. It has a prevalence of about 11% and a remarkable impact on health and costs1. The main phenotypes are CRS with nasal polyps (CRSwNP) and without (CRSsNP)1-3. The majority of western CRS cases are characterized by type 2 -high inflammation with elevated levels of eosinophils, interleukin-4 (IL-4), IL-5 and IL-131. Endoscopic sinus surgery (ESS) has shown to be cost-effective treatment4, if conservative therapy (such as intranasal corticosteroids and nasal saline irrigation) is insufficient1. About a sixth of patients respond unsatisfactory to initial ESS and require revision ESS1.
Early identification of the risk of CRS recurrence after ESS is cost-effective5,6. It helps target treatment correctly39 and prevent permanent tissue changes1. A substantial number of studies have identified risk factors of revision ESS7–15. The studies vary according to sample sizes (n=6615 or n=610009), variable collection (large retrospective data base9 or prospective questionnaires8) or geographic locations (such as US9, Australia16 or Finland7). The commonly recognized risk factors include nasal polyps, asthma, allergy, non-steroidal anti-inflammatory drug (NSAID), exacerbated respiratory disease (NERD) and previous ESS. In a meta-analysis13, the strongest predictors of revision ESS were allergic fungal rhinosinusitis, NERD, asthma, prior polypectomy. However, no prior research has analysed the prediction accuracy of revision ESS at the individual level or for variables having a non-linear association.
The aim of this study was to examine accuracy of personalized prediction of revision ESS, and to identify most important predictor variables via modern machine learning algorithms.