No abstract

Dear editor,We appreciate the opportunity to respond to the comments by Professor Siamak Sabour regarding possible methodological and statistical issues in our article "Ultrasound of pneumothorax in neonates: diagnostic value of the anterior transverse plane and of mirrored ribs." 1 Regarding the diagnostic value of "mirrored ribs," we agree with Prof. Sabour that positive and negative predictive values are influenced by prevalence. 2 Hence, we have calculated other parameters that are not dependent on prevalence: the positive and negative likelihood ratio (LR). Using the data of the study sample on the "mirrored ribs" sign, we find a positive and negative LR of 0.9504 and 1.0290, respectively. In addition, we have added the value of the area under the ROC curve (AUC) as a measure of predictive accuracy, which was 0.5092 (95% confidence interval [CI], 0.4627-0.5556). All these parameters suggest a poor diagnostic accuracy of the mirrored ribs sign, as concluded in our study and in contrast with a prior study. 3 Regarding the diagnostic value of the presence of A-lines in the anterior transverse plane, we calculated a positive and negative LR of >100 and 0.0114, respectively, with an AUC (95% CI) = 0.9869 (0.9774-0.9965), all indicating an excellent diagnostic accuracy. Additionally, we calculated various positive and negative predictive values for different degrees of prevalence, using Bayes' theorem and applying the sensitivity and specificity values obtained in our study (Table 1). The results show that the positive predictive value is not greatly affected by changes in prevalence.Data are openly available in a public repository that issues data sets with DOIs.