We thank Dr. Pickard for pointing out the two errors in our figures. Pregnant women were excluded because the protocol called for venography or angiography in some situations, and these procedures are relatively contraindicated in pregnancy.

Dr. Whitney correctly notes that 5 patients with low pretest probability would be considered to have PE ruled out incorrectly if the negative D-dimer result was relied upon. However, he fails to note that the correct denominator is 521 (that is, <1% of patients would be incorrectly excluded). This number is similar to the negative predictive value of a normal lung scan or angiogram. Dr. Whitney hypothesizes that by not performing follow-up ultrasonography, fatal PE may occur in patients with nondiagnostic lung scans and negative D-dimer results in whom PE was missed. We disagree with this contention because patients with small pulmonary emboli may have spontaneous resolution, but no data exist to prove this either way. Regardless, our strategy is probably better than the standard of care, in which patients with nondiagnostic lung scans are often treated or have PE ruled out on further diagnostic tests (1). We believe that our results show it is worthwhile to do a D-dimer test in combination with clinical pretest probability and that low clinical probability with a negative D-dimer result effectively rules out PE. It is more appropriate to consider that less than 3% of patients with a non-high-probability lung scan and a negative D-dimer result have PE than to consider that 20% of patients who have PE and non-high-probability scans have negative D-dimer test results.

We did not consider smoking, obesity, hypertension, or oral contraceptive use as risk factors because the evidence that these factors are associated with increased risk for PE is not strong.

We may be overdiagnosing PE in patients with DVT, and although we agree that not all patients with DVT have PE, Mr. Egermayer's data are limited by small numbers. However, in most recent clinical trials, the treatment has been the same for PE and DVT. Although there is recent evidence that PE may portend a worse prognosis, there is no evidence that treatment should differ (2). Pulmonary embolism recurrence was not diagnosed retrospectively in our patients. Patients were followed for 3 months and were investigated if symptoms suggestive of recurrence occurred. Finally, adverse events related to therapy are only an issue if patients were inappropriately treated; our study design precluded this possibility.