Health-related quality of life (HRQoL) measurement is considered an important piece of the evaluation process of the burden of a disease from the patient’s perspective (9,13). The COMQ-12 scale is an HRQoL questionnaire for COM originally designed in English in order to assess symptom severity and its relationship with patient’s referred discomfort in their quality of life (16, 17). This scale has been validated in several languages and different socio-cultural backgrounds so that clinical outcomes assessment can be standardized (16). However, the COMQ-12 scale had not been validated in the Spanish language yet. Therefore, standing on previous validation appraisals, we developed a Spanish COMQ-12 version and performed a validation and cultural adaptation process.
Of the 231 patients enrolled in the study, 69.5% of them belong to lower-income levels (Strata 1 and 2) and only 9% of them had attained a professional degree. Consequently, the patients were guided by a trained medical doctor and two otologists through the questionnaire. Even though some previous validations suggest an autonomous completion of the questionnaire (9), there is no general consensus on the most reliable way to answer the COMQ-12 questionnaire (15).
Overall, most of the COMQ-12 items obtained a correlation level superior to 0.8 which means that each item significantly contributes to the global score of the scale. Interestingly, the lowest correlation level was found with the dizziness item, which is considered a less specific symptom of COM (22). On the other hand, the highest correlation level was obtained by the items related to hearing loss, which is linked to limitations in communication that disturbs the quality of life in patients with COM (11).
Internal consistency was assessed using Cronbach’s alpha coefficient, which measures the homogeneity of items on a test (21). Cronbach’s Alpha value for the Spanish version of the COMQ-12 was 0.8526, similar to results published in previous validation studies (16). Table 8 contains a summary of the internal consistency measures obtained with Cronbach’s Alpha for all COMQ-12 versions in different languages. As an additional contribution, we calculated McDonald’s Omega coefficient, which is a robust measurement of internal consistency and has not been used in previous validations of the COMQ-12. McDonald’s Omega value obtained for the overall score was 0.8901, which suggests a high degree of correlation between single items and the global score of the Spanish version of the COMQ-12.
Assessment of test re-tests reliability values remained similar between the first and second visits (15 or 30 days after the first visit). Although a statistically significant increase in the global score was observed, the highest variation of the score was a mean value of 1. ROC curve published in previous validations described a cutoff point between 6 (17), 8 (14), and 9 (18) points to accurately discriminate between COM and control patients. Thus, a difference of one point in the global score in the second visit is not considered relevant in the clinical setting. Furthermore, Lin's Concordance correlation coefficient was 0.945, ranging between 0.931 and 0.96 for all domains, which demonstrates an outstanding test re-test correlation of the Spanish version of the COMQ-12. This value is higher than the value of 0.859 reported in the Dutch COMQ-12 (13) and lower than the Serbian version’s value of 0.985 (9).
On the other hand, factor analysis was performed to asses the internal structure of the Spanish COMQ-12 scale. Two different models were identified according to eigenvalues and scree plot: Model A was obtained with the Exploratory Factor Analysis and two domains were extracted (Figure 1); Model B was established by the Confirmatory Factor Analysis and three domains were defined (Figure 2). Our findings suggested that Model A was similar to the model identified in the Korean validation of the COMQ-12 (17), and Model B was similar to the model proposed in the Serbian validation (9).
Hence, in order to compare the goodness of fit of both models, several statistical tests were conducted. Model B obtained the best indices’ values, which support an acceptable and superior fit of this model. Despite English, Serbian, and Spanish versions of the scale have three domains, we found differences in item distribution and correlation strength indices. Thus, items in the Spanish version of the COMQ-12 are clustered in three domains: “smelly discharge related symptoms” (Q1 and Q2), “hearing loss related symptoms” (Q3-Q7) and “impact on work, lifestyle and health services” (Q8-Q11). Differences from the original English version of the COMQ-12in item distribution and correlation strength indices could be explained by the fact that domains described in the original questionnaire were not developed through factor analysis.
Regarding the concurrent validity of the test, correlation measurements between each item and the visual analogue scale or VAS (Q12) were done. Correlation values between the COMQ-12 global score and VAS were positive and closer to 1 (0.676), and the domain “symptom severity” obtained the higher correlation value ( = 0.608; IC95%= 0.513; 0.689). These findings were similarly described in previously published validations (9)(17)(19). Thus, as the symptom severity score increases, a greater VAS score should be obtained.
Finally, Wilcoxon ranks test addressed the validity of the questionnaire. Statistically significant differences were found in the COMQ-12 global score between COM and control patients (p<0.0001). This finding suggests that the Spanish version of the COMQ-12 can accurately discriminate between COM and healthy patients, similar to Dutch, Serbian, Russian, and Korean versions (8, 13, 14, 16). Likewise, a correlation analysis was also performed classifying COM patients according to their disease activity based on the otoscopic findings. Patients with active disease (active squamous epithelium or cholesteatoma and perforated eardrum with discharge) had a greater overall score with statistically significant differences in “symptom severity” and “lifestyle and work impact” domains. These results are similar to previous COMQ-12 validations (1, 11, 16, 17), and therefore, confirm the relationship between questionnaire score and disease activity.
Strengths and limitations
One of the strengths of this study is the sample size, which has the biggest number of patients with COM compared to previous COMQ-12 validation studies. Also, the otoscopic examination of every patient enrolled was confirmed by two neuro-otologists. Therefore, reliable and homogeneous statistic results were obtained. Likewise, we have addressed our statistical models via exploratory factor analysis, and rigorous statistical strategies for extraction and interpretation were performed. On the other hand, we did not measure the responsiveness of the questionnaire due to surgical intervention. However, this statistical property has not been reported in any previous COMQ-12 validation yet, and we expect to assess it in the second phase of this study.