Stein and colleagues describe a noninvasive management strategy for deciding whether to give thrombolytic therapy to patients who might have acute pulmonary embolism. The authors hypothesize that physicians might hesitate to administer thrombolytic therapy because of the increased risk for bleeding in patients who have received an invasive procedure, namely, pulmonary angiography. To test this hypothesis, the authors might begin by surveying physicians who manage patients with pulmonary embolism to assess why they did not administer thrombolytic therapy. They might find that those physicians did not administer thrombolytic therapy because the indications for thrombolytic therapy are unclear. Although recommendations for thrombolytic therapy have been made, these recommendations are based on soft data and thus are open to question by any practitioner. No survival benefit accrues to treated patients, and complications of treatment are definitely increased. A recent analysis [1] of some of the putative benefits of thrombolytic therapy has suggested that these benefits are illusory.

In their assessment, the authors make several questionable decisions that weight their subsequent analysis of the appropriate management strategy in favor of a noninvasive approach. First, the authors overestimate the risk of angiography. The most important adverse outcome from thrombolytic therapy is intracerebral hemorrhage. According to the data that the authors cite from the Thrombolysis in Myocardial Infarction trial [2], the frequency of intracerebral hemorrhage was not altered by an invasive compared with a noninvasive management style. In patients who received earlier angiography, the excess complications were local bleeding. Even this risk is probably magnified because the patients had arterial rather than venous access and thus were probably at greater risk for large hematomas. The risk of right-heart catheterization is also probably overstated. After analyzing data from the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study, the authors cite a risk of 1.3% for serious complications. However, that article [3] indicates that most of the serious complications were believed to be caused by the morbid condition of the patient at the time of angiography and not by the angiographic procedure itself.

Second, the authors underestimate the risk of pursuing a noninvasive strategy. In addition to the risk of anticoagulation, the patient who does not have pulmonary embolism and is treated with thrombolytic therapy without a firm diagnosis of pulmonary embolism will be labeled with this diagnosis in the future. Thus, an unknown but substantial number of other interventions may occur on the basis of this previous erroneous diagnosis of pulmonary embolism. These subsequent interventions could be associated with substantial morbidity or mortality.

In summary, the authors provide no cogent evidence to support the strategy that they espouse. In fact, an invasive strategy directed at making a diagnosis of pulmonary embolism and subsequent placement of an inferior vena cava filter would be better. Patients who survive the initial embolism but eventually die of embolism do so because of recurrent embolism [4]. An inferior vena cava filter is the only intervention that immediately reduces the risk for recurrent embolism [5]. Anecdotal reports suggest that thrombolytic therapy might even induce recurrent embolism in some patients.