72 Chapter 4 Reliability Internal consistency: Cronbach’s alpha for the impact items of the OQUA was .916, indicating strong evidence for a good internal consistency. Test-retest reliability: The mean interval between T0 and T1 was 8.69 days (range, 6-14 days). The quadratic weighted kappa’s with associated 95% confidence interval bounds in the parentheses and standard error of measurements of the 26 items, 9 scored a very good (agreement kappa (ak) > 0.80) test-retest reliability. Sixteen items scored good (ak = 0.6-0.8), and only one item (hearing loss Q1; ‘Can you hear someone approaching from behind?’) scored moderate (ak < 0.6). Four of the items had a very good test-retest reliability (ICC > 0.8), and the other five had a good test-retest reliability (ICC = 0.6-0.8). Responsiveness: Change scores on both the OQUA impact items and GHSI total score were normally distributed. Responsiveness showed good correlation between OQUA impact and GHSI using Pearson’s coefficient (−.711). This result matches our hypothesis. Correlation between the GRS and the complaints of the OQUA was calculated using the Spearman ρ and showed (.623). Tables 3 and 4 show the responsiveness results in a subgroup analysis of cholesteatoma and otosclerosis patients. No other subgroup analysis was made due to small sample size. In the both subgroups, the preoperative response of the OQUA was spread over multiple complaints. The overall impact preoperative response was higher than postoperative response, but the change in complaints varies from patient to patient. Sequence: A considerable number of patients made a comment about the sequence of the questions. In the first version of the OQUA, 5 the sequence of the items was more randomly distributed to avoid duplication of answer categories to items that appeared to be alike. Keeping these questions close together could have interfered with the explorative factor analysis. An example of an item which appeared to be alike is ‘I have ear discharge’ compared to ‘Liquid comes out of my ear’. After finishing the validation and item reduction, we have changed the sequence of the items in order to be more logical and bringing the items about the same domain together. In Figure 1, the latest version (2020) of the OQUA is presented.
RkJQdWJsaXNoZXIy MjY0ODMw