Errors in Questionnaire Scoring Algorithm

In the Original Investigation titled "Development and Initial Validation of a Consumer Questionnaire to Predict the Presence of Ear Disease," published online on August 3, 2017, and in the October 2017 issue of JAMA Otolaryngology–Head & Neck Surgery, there were errors in the analysis code for 3 of the questionnaire items (questions 16, 18, and 19). For these 3 questions, the 4-item rating scales were dichotomized incorrectly. Using the corrected algorithm resulted in changed sensitivity and specificity, odds ratios, and 95% CIs, as detailed herein. The corrections are also explained in a Letter to the Editor. In the abstract Results, descriptions of the sensitivity and specificity have been corrected to 90% (95% CI, 84%-94%) and 72% (95% CI, 59%-82%), respectively, for the training sample and 76% (95% CI, 61%-87%) and 80% (95% CI, 51%-95%) for the test sample, respectively. In the Statistical Analysis paragraph of the Methods section, rating scale items should have been dichotomized as 0 for "never" and "occasionally" and as 1 for "sometimes" and "always" rather than scored on a scale of 0 to 4. In the Results section, the odds ratio and 95% CI for the association between CEDRA score and presence of disease changed. In addition, the probability, sensitivity, and specificity values changed. In the third paragraph of the Discussion, all percentages have changed. Figure 2 has been replaced with a graph incorporating a corrected receiver operating characteristic curve, and the figure caption has been updated. This article was corrected online.

https://ift.tt/2FDFfqG