Clinical and symptom criteria for the accurate diagnosis of chronic rhinosinusitis

Abstract

Objectives/hypothesis: The objectives of this prospective, double-blind diagnostic cohort study were to explore the relationship between patients' reported symptoms of chronic rhinosinusitis (CRS) and radiographic findings in CRS to define clinically based diagnostic criteria for CRS and to determine a classification scheme that would allow for the accurate diagnosis of patients with CRS.

Methods: A prospective series of adult patients undergoing evaluation for potential chronic rhinosinusitis was studied. Patients' symptoms were tabulated with the Rhinosinusitis Symptom Inventory (RSI), which records symptoms associated with CRS on a Likert scale (5 = maximally severe). Medication and resource utilization were also tabulated. Scores for nasal, facial, oropharyngeal, systemic, and total symptom domains were computed (100 = maximum severity). Concurrent paranasal sinus computed tomography was obtained for each patient and scored according to the Lund-McKay system with reviewer blinding. Patients were classified as normal or representing true CRS according to previously published diagnostic radiographic criteria under two analyses. In analysis 1, Lund scores < or = 1 were considered nondiseased and scores > or = 2 were considered as representing true CRS. In analysis 2, Lund scores < or = 3 represented normal and scores > or = 4 represented true CRS. For each analysis, univariate statistics were computed to determine differences among patients' reported symptoms with and without true CRS. Subsequently, classification analysis using classification regression trees was conducted to determine heuristics among clinically based criteria for the diagnosis of CRS.

Results: Seven hundred three adult patients were enrolled (mean age, 43.1 years). The most commonly reported symptoms in order of decreasing severity and presence were nasal obstruction, nasal congestion, discharge, fatigue, headache, facial pressure, and dysosmia, all with mean severity scores > or = 2.0. Patients were previously treated for approximately 3 months each with topical nasal steroids, oral antihistamines, and oral antibiotics. Patients missed substantial workdays (mean, 3.8/year) and incurred significant physician visits (mean, 3.5/year) as a result of CRS. The nasal symptom domain exhibited the greatest severity (51.8) followed closely by the facial symptom domain (47.3). According to the analysis 1, 144 (20.5%) patients were classified as normal and 559 (79.5%) as true CRS. Only the presence of polyps (P < .001) and dysosmia (P = .008) distinguished between normal and diseased patients; there were no significant differences between groups for the other symptoms and RSI domains. Classification analysis revealed that the presence of polyps, absence of dental pain, and low congestion/obstruction scores in the presence of dental pain predicted true CRS. Finally, a history of extended antibiotic courses before presentation also predicted true CRS. In the second analysis, 243 (34.6%) were normal versus 460 (65.4%) with true CRS. Only polyp presence and dysosmia exhibited statistically significant differences between normal and diseased patients (P < .001). The mean RSI nasal domain was higher for true patients with CRS (P = .01). Classification analysis identified polyp presence and dysosmia score >2 effectively predicted CRS. Fatigue < or = 2, discharge < or = 4, and obstruction < or = 3 further predicted for CRS. Symptoms despite a > or = 5.5-week trial of nasal steroids also predicted CRS. Other major and minor symptoms, RSI domains, and resource factors failed to influence diagnostic classification.

Conclusions: The diagnosis of CRS based on symptom criteria is difficult because most symptoms (other than dysosmia) do not distinguish between radiographically normal and diseased patients. A classification scheme for CRS based on the presence or absence of polyps is valid. It is possible to classify patients with true CRS based on symptoms, but this requires complicated heuristics. Corroborating radiographic evidence should be required to establish the diagnosis of CRS before long periods of therapy or surgical intervention.