We appreciate the thoughtful letters about our article [1]. Although a difference of opinion appears to exist, our article may have been misinterpreted. We did not imply that we favor a broader use of thrombolytic drugs for patients with acute pulmonary embolism. Our purpose was to assess the relative risks of thrombolytic therapy administered on the basis of an invasive compared with a noninvasive diagnosis. The physician must determine the indications for thrombolytic therapy.

The indications that we use, hypotension or hypoxia while the patient is receiving high doses of oxygen or echocardiographic evidence of right ventricular failure, are guidelines that assist in identifying patients who may not survive the acute episode of pulmonary embolism. Perhaps unstable patients may do better with open embolectomy, catheter-tip embolectomy, or catheter-tip fragmentation than with thrombolytic therapy. Regardless, if a physician selects thrombolytic therapy as the treatment of choice, our data allow assessment of the risks for major complications of thrombolytic therapy administered on the basis of an invasive or noninvasive diagnosis.

Clearly, both a strong impression of acute pulmonary embolism and a clear indication for thrombolytic therapy must exist. If the patient requires a lifesaving intervention and if the physician selects thrombolytic therapy as the intervention, then our data suggest that fewer major complications will occur with thrombolytic therapy that is administered on the basis of a noninvasive diagnosis (even if the likelihood of pulmonary embolism is relatively low). We are not suggesting a more liberal use of thrombolytic therapy, nor are we recommending a strategy for its use. Rather, we are evaluating the risks for different approaches to the diagnosis if thrombolytic therapy is necessary. If a patient is likely to survive after an angiographic diagnosis and placement of an inferior vena cave filter, then thrombolytic therapy would not be necessary. We agree that it is important to establish a correct diagnosis for future care; patients should not be labeled incorrectly. Perhaps noninvasive investigations such as cardiac echocardiography and noninvasive leg tests could be used with ventilation-perfusion lung scans to assist with making a diagnosis.

We agree that intracerebral hemorrhage is the worst complication of thrombolytic therapy. However, we also believe that—in this day of transfusion-induced illness—bleeding severe enough to require 2 or more units of blood is a major complication. We agree that the relative risk for bleeding from thrombolytic therapy after percutaneous entrance of a vein may be different than that after percutaneous entrance of an artery. That is why we did sensitivity analyses and made calculations at the lower end of the 95% CI for assessing the relative risks for bleeding on the basis of data from coronary arteriography. We even went beyond the 95% CI in the sensitivity analysis. If major bleeding with thrombolytic therapy after pulmonary angiography is only 1% more than major bleeding with a noninvasive diagnosis (that is, 5.2% compared with 4.2%), calculations showed that it would be safer to treat patients on a noninvasive basis if the probability of pulmonary embolism was more than 55%. This does not mean that we are suggesting a random use of thrombolytic therapy. By using our data, physicians may be able to make an informed judgment about the risks of therapy.

Unfortunately, major bleeding at the site of insertion of the catheter in the femoral vein cannot always be controlled by pressure. We interpret the loss of 2 or more units of blood at this site as a major complication. All reported investigations of thrombolytic therapy in patients with pulmonary embolism excluded patients who had recent surgery, blood dyscrasia, severe hepatic disease, or renal disease. Therefore, the reported frequency of bleeding in all reported studies applies to patients who had a low risk for bleeding.

Regarding the risks for serious complications after pulmonary angiography, on the basis of data from PIOPED, most serious complications developed in patients who were sent to have angiography from the intensive care unit and, therefore, were in serious condition [2]. The complications, however, were caused by pulmonary angiography and not by the underlying morbid conditions.