The corresponding author sent e-mail to Dr. John Phillip, the original author of COMQ-12, and obtained permission and received the original version of COMQ-12.

The original English version of COMQ-12 consists of 12 questions. Each item is scored from 0 to 5, the higher number the more discomfort it means. The original COMQ-12 was translated from English to Korean by bilingual otolaryngologist in English and Korean. Next, the translation was checked by a person who majored in Korean literature. Then the naive Korean version was translated back to English. Both the original and retranslated English version of the COMQ-12 were compared by a person who majored in English literature and suggested minor changes (Appendix 1).

Power analysis were performed to estimate the number of subjects (effect size 0.28, a-risk 0.05). In total, 212 individuals were enrolled in this study. One hundred and six patients who visited Yonsei University Wonju Severance Christian Hospital and diagnosed COM and planned surgery were enrolled in the study. To compare with control group, 106 healthy volunteers who have not experienced COM were asked randomly from Yonsei University Wonju Severance Christian Hospital.

All subjects participated in informed voluntary consent, and were independently answered each question of K-COMQ-12. In case of COM patients, they answered to K-COMQ-12 before they took surgery. There was no reported difficulties answering each question, and participants received no help in answering questions. Patients who could not read and understand the Korean-language were not enrolled in this study. Demographic informations of subjects such as patient’s sex, age and body mass index (BMI) were collected. Informations of medical history, including hypertension, diabetes mellitus, lung or heart disease, cerebrovascular accident, chronic kidney disease, liver disease and frequency of noise exposure were also collected.

The Institutional Review Board of Yonsei University Wonju Severance Christian Hospital approved this study (CR316063).

Statistical analysis was conducted using the Statistical Package for Social Sciences Software (SPSS 22.0 IBM Corp., Armonk, NY, USA). Cronbach’s alpha and item-total correlation were measured to evaluate internal consistency. We used Kaiser-Meyer-Olkin (KMO) test to measure sampling adequacy. KMO value is considered ‘good’ if it is greater than 0.70, and ‘excellent’ if it is greater than 0.90. Generally, it is known that factor analysis is appropriate if the KMO value is greater than 0.7. We also performed factor analysis, and the number of factors was determined using the eigenvalue and scree plot curve. Varimax rotation was used to simplify the interpretation of data. The independent t-test and one-way analysis of variance with Bonferroni method for continuous variables, chi-square or Fisher’s exact test for categorical variables were performed to construct validity. To facilitate the clinical interpretation of K-COMQ-12 scores, a threshold score for distinguishing normal from patients of COM was derived from the receiver operating characteristic (ROC) curve. We assessed significant differences when the p-value was less than 0.05 (p＜0.05).

For the K-COMQ-12, Cronbach’s alpha was 0.939, suggesting good internal consistency. We also assessed item-total correlation coefficients. It ranged from 0.548 to 0.822 that indicates all individual items correlated well with the total score (Table 1).

As the KMO value was 0.884, factor analysis is appropriate for our data. Factor analysis is generally used to group items of questionnaire with similar features. We determined the 2 factors were extracted using eigenvalues and a scree plot curve (Fig. 1). They explained 60.26% and 7.33% of all items of K-COMQ-12. The factor 1 includes 9 items (item 3, 4, 5, 6, 7, 8, 10, 11, 12), and the factor 2 includes 3 items (item 1, 2, 9) (Table 2). However, we could not find appropriate titles covering the items that make up each factor.

The item-communality values, which refers to the amount of variance of an item as predicted by the loading factor, ranged from 0.451 to 0.787 (Table 3). Generally, it is considered that each item loads well on factor when the value of communality is greater than 0.45.

We assessed the construct validity of K-COMQ-12 by comparing scores from COM patients to those of the control group. Among total 212 subjects which consist of 106 COM group and 106 normal group, clinical baseline characteristics of each group are shown in Table 4. There were no significant differences in many factors including subject’s age, sex, BMI and their medical histories. 34.9% of COM patients were performed previous ear surgery, 14.2% of COM patients had experience of noise exposure, and 11.3% had or had been wearing hearing aids in the past. They were statistically significant when comparing normal group (Table 4).

The control group showed 1.99 of mean K-COMQ-12 score [4.34 of standard deviation (SD)] whereas the COM group showed 22.69 (12.41 of SD), and they were statistically significant (Table 5).

We classified the severity of COM according to the surgery which the patient decided to undergo. Typically, long-lasted infection with poor Eustachian tube function induces relatively severe COM, and usually requires mastoidectomy, however, tympanoplasty only is often sufficient for COM with relatively intact eustachian tube function and relatively short-lasting COM [17]. Following the type of surgery, 106 COM subjects were divided into two groups; the ‘Tympanoplasty only’ group and the ‘Mastoidectomy’ group. The ‘Tympanoplasty only’ group was defined as COM patients who underwent only tympanoplasty without mastoidectomy after an evaluation by otolaryngologist, and ‘Mastoidecotmy’ group was defined as COM patients who decided to undergo mastoidectomy regardless of whether tympanoplasty was performed or not. Among 106 of COM subjects, 38 were tympanoplasty only group and 68 were mastoidectomy group. The score of K-COMQ-12 showed significant differences among control, ‘Tympanoplasty only’ group and ‘Mastoidectomy’ group (p＜0.001). However, in the post-hoc test, the difference in the K-COMQ-12 score between the ‘Tympanoplasty only’ group and the ‘Mastoidectoy group’ was not statistically significant (p=0.700).

We tried to classify the COM subjects by the disease activity of COM according to the physical examination. ‘Inactive COM’ group was defined as having no evidence of active infection with dried up tympanic membrane (TM) whether the perforation is or not on physical examination. ‘Active COM’ group was defined as showing wet TM or retracted or adhered TM to middle ear space. We also compared the score of K-COMQ-12 among the control, ‘Inctive COM’, and ‘Active COM’ group. The difference of K-COMQ-12 score according to the disease activity of COM based on physical examination were statistically significant among all three groups (p＜0.001). The disease activity of COM goes higher, the K-COMQ-12 score went higher (Table 6).

To assess the meaningful cut-off value between COM patients and normal group, the ROC curve was used (Fig. 2). The value of area under the curve was 0.969, which implies very high discriminatory power to distinguish subjects with the control and COM patients (p＜0.001). We could also identify which cut-off point that indicates best balance between sensitivity and specificity using ROC curve. A cut-off score of 6 of K-COMQ-12 sets 0.915 of sensitivity and 0.934 of specificity. Therefore, using the K-COMQ-12 score of 6 of threshold, 91.5% of patients were correctly classified as COM, and 93.4% of non-patients were correctly categorized as normal.