Rapid on-site cytologic evaluation during endobronchial ultrasound-guided transbronchial biopsy

摘要

With interest, we have read the article by Lin et al., a publication of the Journal of the Formosan Medical As- sociation of November 2020. 1 Although interesting, there exist some potential statistical limitations, which could markedly affect the primary message of the study. Diagnostic accuracy studies should report appropriate estimates when comparing between different diagnostic tests. When investigating the diagnostic accuracy of rapid on-site cytologic evaluation (ROSE) with reference to different standard, sensitivity, specificity, positive/ negative predictive and diagnostic accuracy were re- ported only. It should be noted that the so-called diag- nostic accuracy [(true positives t true negatives)/(true positives t false negatives t false positives t true negatives)] is a “rough” estimate of the accuracy of an index test, since false positives (a reflection of speci- ficity) and false negatives (a reflection of sensitivity) are assumed to have the same importance. 2 An extreme condition is: test A with 100% specificity and 0% sensi- tivity and test B with 0% specificity and 100% sensitivity would have equal diagnostic accuracy for a 50% occur- rence condition. Therefore, other measure, such as diagnostic odds ratio, negative-to-positive likelihood ratio, diagnostic added value with additional receiver operating characteristic curves should also be taken into consideration. Moreover, accuracy measures could be admissible, while the diagnostic added values might still be negligible clinically.