Lung scanning for pulmonary embolism: clinical and pulmonary angiographic correlations

A group of 78 patients with suspected pulmonary embolism was studied by both ventilation perfusion lung scanning and pulmonary angiography. Symptoms and clinical signs were analysed using Bayesian techniques to produce pre-test odds for pulmonary embolism in individual patients. While, as a group, those with embolism could be discriminated from those without on this basis, major overlap existed between the groups, invalidating the use of this approach for individual patients. Strict diagnostic criteria for interpretation of lung scans were accurate using pulmonary angiography as the 'gold standard', but at the expense of a significant number of patients (38 per cent) in the indeterminate (non-diagnostic) group. In the 48 patients in whom the test yielded a diagnostic result, there was a sensitivity of 100 per cent (15/15) and a specificity of 97 per cent (32/33). In the series as a whole, the likelihood of lung scanning correctly diagnosing pulmonary embolism was 55 per cent (15/27) and of correctly excluding embolism, 63 per cent (32/51). By the use of strict criteria for interpretation of lung scanning, reliable information can be obtained on the presence or absence of pulmonary embolism in a large proportion of patients suspected of having the condition. Such information is more discriminating than clinical signs and symptoms.