A way to select on clinical grounds patients with high risk for pulmonary embolism: a retrospective analysis in a nested case-control study.

We studied 196 patients with suspicion of pulmonary embolism (PE), subsequently confirmed in 98 by positive pulmonary angiography and excluded in 98 by normal or near-normal perfusion lung scan. Patients had a clinical questionnaire for history, and, soon after. a radiograph, blood gas analysis, and an ECG. Clinical and instrumental signs were matched in patients with confirmed and unconfirmed PE to find those more frequent in embolic patients and, thus, more characteristic of PE. The following were: previous PE, immobilization and thrombophlebitis (p < 0.05); dyspnea and cough (p < 0.05); enlarged descending pulmonary artery (DPA), enlarged right heart, pulmonary infarction, Westermark sign (p < 0.001), and elevated diaphragm (p < 0.05); hypoxemia. No ECG sign was more frequent in PE. Thereafter, all variables were processed separately with a logistic multiple regression analysis and those significantly associated to PE were tested in a final logistic model that was able to predict the actual result of angiography or scintigraphy; accordingly, previous PE, immobilization, thrombophlebitis, enlarged DPA, pulmonary infarction, Westermark sign, hypoxemia were significantly associated with a high risk of PE (from 2.8 to 15 times greater than in patients without these signs). Therefore, we may conclude that clinical assessment and noninvasive tests may help to detect patients at higher risk for PE where heparin coverage should be started while waiting for conclusive diagnostic procedures.