The results of the study by Flinn et al1 reemphasize that the primary and secondary prevention of pulmonary embolism in a high-risk population may reduce adverse outcomes. We were impressed by the authors' 94% accuracy rate for venous duplex ultrasound scanning. This is not a common experience in the asymptomatic deep vein thrombosis (DVT) population. Pooled data from studies using venography (the "gold standard") systematically show a 79% sensitivity using venous duplex ultrasound scanning for proximal DVT and a positive predictive value of 69%,2 values that drop in patients with craniotomy to 38% and 56%, respectively.3 The authors' accuracy rate was derived from "an extended period of experience,"4 with no mention of the number of patients or the method of evaluation used, which leaves the readers to surmise its validity. When the authors quote "accuracy" rate, are they referring to the positive predictive value?