Neither Dr. Merrell and colleagues nor Dr. White challenges our conclusion that color Doppler ultrasound examinations are insensitive to proximal deep venous thrombosis in asymptomatic high-risk patients after hip or knee arthroplasty [1].

Dr. Merrell and colleagues question our selection of only seven sites from the larger RD Heparin Arthroplasty Group trial that met the prospective entry criteria. The LDS Hospital peripheral vascular laboratory used gray-scale imaging and compression (not color Doppler); therefore, data from their center and other centers not using color Doppler ultrasound were excluded. Results from duplex scans at their center are the subject of another article [2], which notes that the 100% sensitivity of duplex scanning for detection of four proximal clots has a 95% CI that includes a sensitivity as low as 40%. The centers included did use comparable (although not identical) equipment in a consistent fashion. Finally, Merrell and colleagues misstate our conclusion as applying to duplex scanning, when in fact we discussed color Doppler ultrasound instead.

As for Dr. White's questions, first, we had no reason to question the diligence of the technicians. In the absence of evidence for this hypothesis, he asserts that our frequencies of false-positive and false-negative results were higher than those in other unspecified studies of color Doppler ultrasound. We are unaware of similar studies of sound method with significant patient numbers. Second, although use of an expert panel might have changed certain data, it is not clear that it would have changed study outcome. Moreover, expert panel readings are not typical of patient care at most hospitals. Our methods more closely mirrored contemporary medical practice, in which a local radiologist reports a reading. Third, the 8% false-positive rate (defined as 100% –specificity) is not as atypical as Dr. White believes. Both studies he cites included fewer than half the number of patients and reported data in legs rather than patients studied. In addition, the 95% CI for the false-positive rate in one study (3.1%; 95% CI, 1.5% to 7.5%) he cites is close to including 8% [3]. Two studies we cited [4, 5] reported false-positive rates similar to our own (6% and 3%, respectively; the 95% CIs for these observations include 19% and 15%, respectively). These data relate to patients with asymptomatic joint replacement.

Dr. White closes as we did, by noting that studies like ours cannot define the clinical significance of the proximal clots missed (or detected) by ultrasound techniques. We are unaware of data showing that proximal clots of less than 4 cm and those "barely extending into the popliteal vein" are clinically unimportant. We reported that one of four occlusive proximal clots found by venography was missed by color Doppler ultrasound. How many patients to work up with venograms, how to treat, with what drugs, how intensively, and for how long remain important clinical questions. We still conclude that color Doppler ultrasound is no shortcut to the answers.