Premium
Effect of symptom‐based risk stratification on the costs of managing patients with chronic rhinosinusitis symptoms
Author(s) -
Tan Bruce K.,
Lu Guanning,
Kwasny Mary J.,
Hsueh Wayne D.,
ShintaniSmith Stephanie,
Conley David B.,
Chandra Rakesh K.,
Kern Robert C.,
Leung Randy
Publication year - 2013
Publication title -
international forum of allergy and rhinology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.503
H-Index - 46
eISSN - 2042-6984
pISSN - 2042-6976
DOI - 10.1002/alr.21208
Subject(s) - medicine , chronic rhinosinusitis , risk stratification , stratification (seeds) , intensive care medicine , sinusitis , surgery , seed dormancy , botany , germination , dormancy , biology
Background Current symptom criteria poorly predict a diagnosis of chronic rhinosinusitis (CRS) resulting in excessive treatment of patients with presumed CRS. The objective of this study was analyze the positive predictive value of individual symptoms, or symptoms in combination, in patients with CRS symptoms and examine the costs of the subsequent diagnostic algorithm using a decision tree–based cost analysis. Methods We analyzed previously collected patient‐reported symptoms from a cross‐sectional study of patients who had received a computed tomography (CT) scan of their sinuses at a tertiary care otolaryngology clinic for evaluation of CRS symptoms to calculate the positive predictive value of individual symptoms. Classification and regression tree (CART) analysis then optimized combinations of symptoms and thresholds to identify CRS patients. The calculated positive predictive values were applied to a previously developed decision tree that compared an upfront CT (uCT) algorithm against an empiric medical therapy (EMT) algorithm with further analysis that considered the availability of point of care (POC) imaging. Results The positive predictive value of individual symptoms ranged from 0.21 for patients reporting forehead pain and to 0.69 for patients reporting hyposmia. The CART model constructed a dichotomous model based on forehead pain, maxillary pain, hyposmia, nasal discharge, and facial pain (C‐statistic 0.83). If POC CT were available, median costs ($64‐$415) favored using the upfront CT for all individual symptoms. If POC CT was unavailable, median costs favored uCT for most symptoms except intercanthal pain (−$15), hyposmia (−$100), and discolored nasal discharge (−$24), although these symptoms became equivocal on cost sensitivity analysis. The three‐tiered CART model could subcategorize patients into tiers where uCT was always favorable (median costs: $332‐$504) and others for which EMT was always favorable (median costs −$121 to −$275). The uCT algorithm was always more costly if the nasal endoscopy was positive. Conclusion Among patients with classic CRS symptoms, the frequency of individual symptoms varied the likelihood of a CRS diagnosis marginally. Only hyposmia, the absence of facial pain, and discolored discharge sufficiently increased the likelihood of diagnosis to potentially make EMT less costly. The development of an evidence‐based, multisymptom‐based risk stratification model could substantially affect the management costs of the subsequent diagnostic algorithm.