The article by Wisnivesky et al,1 and its prediction rule to accurately identify and isolate patients with active tuberculosis (TB), outlines an interesting set of criteria. The authors, however, mistakenly enrolled in their study only patients who were actually isolated; they should have included the patients who had active TB but were not initially isolated. Excluding those patients who should have been isolated resulted in losing a significant number of patients (8/64 [12.5%]) from the typical population we aim at isolating to avoid nosocomial infection. These excluded patients might have had different characteristics from the patients actually included in the study; therefore, the prediction rule Wisnivesky et al1 propose may have different consequences. In fact, in our institution 8 (33.3%) of 24 patients who were not initially isolated but later were found to have active TB had shortness of breath (unpublished data), which is claimed to be a negative predictor in the study.1