Thirty out of this study population have been previously reported and this is add-on study to previous publication that correlated preoperative findings on computed tomography (CT) with poor olfactory outcome [11]. Fifty subjects, who were diagnosed with CRS and NP with a smelling disturbance and were treated with ESS, were enrolled in this study.

Frontoethmoidectomy with middle meatal antrostomy (with/without sphenoidectomy) was performed in all cases. Surgery in the olfactory cleft and middle turbinectomy were not performed. Preoperative steroids were not given and nasal steroid spray or antibiotics were used on the all subjects for 1-3 months after surgery.

The diagnosis was based on at least a 3-month history of rhinosinusitis, findings of nasal endoscopy, CT findings, and olfactory function tests. Patients were excluded if they had a history of head trauma, acute olfactory dysfunction after previous upper respiratory tract infection, age <15 or >65 years, history of previous sinus surgery, and combined septoturbinoplasty. All patients filled out questionnaires and underwent the butanol threshold test (BTT) and the cross-cultural smell identification test (CCSIT) at the initial preoperative visit and at 6 months after surgery.

An endoscopic examination had been performed by one investigator blinded to study design before postoperative olfactory tests were conducted. Pathological findings, including discharge, edema, and polyp in the nasal cavity were checked. According to the Lund-Kennedy score [12], subjects were divided into two groups: favorable wound healing group, who was scored 0, and unfavorable wound healing group, who showed pathologic findings. Unfavorable group was divided according to score: mild group who was scored 1-6 and severe group who was scored 7-12. This study was approved by the Institutional Review Board of Gyeongsang National University Hospital.

The preoperative ostiomeatal unit (OMU) CT images at 2.5 mm thickness were evaluated in the axial and coronal planes without contrast enhancement. All of the paranasal sinuses were graded on a scale of 0-2 according to the Lund-Mackay system [13]. CT grading of olfactory cleft opacification was done as previously reported [11]. CT grading was performed by one investigator blinded to this study design. The borders of the olfactory cleft were defined as follows. The anterior, posterior, medial, and lateral borders were the anterior attachment of the middle turbinate, anterior wall of the sphenoid sinus, nasal septum, and middle and superior turbinates, respectively. The superior border was the skull base. The inferior border was the inferior portion of the middle turbinate. We divided the olfactory cleft into two parts: anterior olfactory cleft and posterior olfactory cleft, which were split by the anterior end of the superior turbinate. In addition, the olfactory cleft was divided into upper and lower parts, which were split by imaginary midline (Fig. 1). The adoption of the grading of upper part of the olfactory cleft was because these areas are the true sensory cavity [14]. Each olfactory cleft opacification was graded on a scale of 0-4 via the ratio of the opacified area to the whole area of the olfactory cleft (0, no opacification; 1, ≤25%; 2, 25%-50%; 3, 50%-75%; 4, >75%).

The odor threshold was determined using the BTT. Four percent N-butanol (Sigma-Aldrich, St Louis, MO, USA) was serially diluted 3-fold with mineral oil (Sigma-Aldrich) to form a series of 10 concentrations. The test was performed in each nostril separately. The examinees were given two polyethylene bottles, one with mineral oil and the other with butanol. The patients were asked to choose the bottle with butanol. The test was repeated until the butanol bottle was correctively identified by the examinee in five consecutive trials. The examination was started at concentration level of 10. The concentration level was designated as the threshold level. For the BTT, anosmia was defined as concentration levels of 0-1, severe hyposmia as levels 2-3, moderate hyposmia as levels 4-5, mild hyposmia as levels 6-7, and normosmia as levels 8-9 [15].

To test odor identification, the CCSIT (Sensonics, Haddon Heights, NJ, USA) was performed in both nostrils simultaneously. The threshold of the CCSIT was defined as the mean odor threshold of both nostrils. Anosmia was defined as concentration levels 0-3, severe hyposmia as levels 4-5, moderate hyposmia as levels 6-8, mild hyposmia as levels 9-10, and normosmia as levels 11-12 [16].

The validated questionnaire including 14 odors common in Korea was performed in the presence of a trained research coordinator. Patients were asked to grade their olfactory dysfunction on a scale of 0-4, with 0 being a normal sense of smell to 4 being complete anosmia. The correlation between the responses to the questionnaire and objective tests was validated as described previously [17]. Postoperative success rate was classified as complete success (improvement of all olfactory tests), partial success (improvement of at least two olfactory tests), and others (failure).

Statistical analyses were performed with SPSS ver. 18.0 (SPSS Inc., Chicago, IL, USA). The Student's t-test (or Mann-Whitney U test) or χ²-test (or Fisher's exact test) were applied to determine statistical differences (P<0.05). Data was expressed as the mean±standard deviation (SD).

There were 31 male and 19 female patients. The ages ranged from 16 to 63 years (mean, 40.8 years). Preoperative questionnaire, BTT and CCSIT scores averaged 12.06±12.18, 3.73±2.90, and 5.39±2.78, respectively. Preoperative Lund-Mackay score, anterior olfactory cleft score (AOCS), posterior olfactory cleft score (POCS), upper AOCS, and upper POCS was 16.82±5.17, 4.29±3.08, 4.18±3.15, 3.95±3.14, and 4.30±3.17, respectively. In terms of postoperative wound healing status, demographic data and associated medical history are shown in Table 1. There were no differences in demographics, preoperative olfactory tests, and preoperative CT findings.

For favorable wound healing group (n=30), the mean improvement of questionnaire, BTT, and CCSIT were 19.70±14.31, 2.88±2.55, and 2.24±2.74, respectively. Improvement rates of questionnaire, BTT and CCSIT were 90%, 83%, and 63%, respectively. All patients with improved BTT also showed improved questionnaire responses. Among the 20 patients with improved CCSIT, postoperative BTT and questionnaire improved in 19 patients. There was only one case of isolated improvement of CCSIT. Nineteen out of 30 patients (63%) showed complete success, followed by 20% partial success (n=6). Five subjects (16%) were included in the failure criteria.

For the unfavorable wound healing group (n=20), the mean improvement of questionnaire, BTT, and CCSIT scores were 11.38±17.06, 1.23±3.38, and 0.62±3.25, respectively. Improvement rates of questionnaire, BTT, and CCSIT were 60%, 45%, and 25%, respectively. Twelve out of 20 patients (60%) were classified as failure. Five subjects (25%) and three subjects (15%) were classified as complete success and partial success, respectively. Subjects in the favorable wound healing group showed more marked mean improvement of olfactory results, improvement and success rates than the unfavorable wound healing group. Mild group showed higher mean improvement of olfactory results than severe group in the unfavorable group (Table 2).

In the favorable wound healing group, patients with complete success had higher mean score of preoperative olfactory tests than those with partial success or failure. There was a significant difference in the preoperative questionnaire (Table 3) (P=0.005). In preoperative CT, Lund-Mackay score showed no differences between the two groups. However, there were significant differences in olfactory cleft scores, such as AOCS (P=0.044), upper AOCS (P=0.025), and upper POCS (P=0.032), between the two groups (Table 3).