Diagnosing pulmonary embolism using clinical findings

To determine the value of clinical findings in the diagnosis of pulmonary embolism, we analyzed the records of 82 patients from three different hospitals who underwent pulmonary arteriography to rule out pulmonary embolism. We recorded 92 items of clinical information, including lung ventilation-perfusion scan results, for each patient. Although the diagnostic power of any single variable was marginal, an eight-item decision rule derived using discriminant analysis correctly predicted the outcome of arteriography in 82% of the cases. When tested on a different group of 68 patients from four hospitals, the rule accurately identified patients with low or high likelihood of having positive arteriograms. Clinical use of such a rule could reduce the need for arteriography in this population and would expedite decisions about anticoagulant therapy and further diagnostic testing.