The Diagnostic Approach to Pulmonary Embolism
Deep venous thrombosis and pulmonary embolism both present diagnostic problems1,2. Both disorders have a "gold-standard" of diagnosis, i.e., contrast venography for DVT, and pulmonary arteriography for pulmonary embolus, though both require careful performance and interpretation by experienced observers. Since venography and pulmonary arteriography are invasive, relatively expensive, and not universally available, diagnostic strategies have been developed utilizing non-invasive tests.
The classical signs and symptoms of deep vein thrombosis include pain, swelling, redness, and superficial warmth of the affected area. Deep venous thrombosis cannot be reliably diagnosed based on clinical signs and symptoms, because in more than half these patients, the diagnosis cannot be confirmed by venography. Validated noninvasive diagnostic techniques include iodofibrinogen leg scanning, impedance plethysmography, Doppler examination, B-mode ultrasonography (the combination of Doppler examination with ultrasonography is called "Duplex" exam). Currently, the two most common methods of diagnosis employed in clinical practice are impedance plethysmography and compression ultrasonography, because they are accurate and non-invasive.3 Wheeler, summarizing 25 studies of ultrasonography and 18 studies of impedance plethysmography in the diagnosis of symptomatic proximal deep vein thrombosis, reported that the sensitivity and specificity of ultrasonography (96% and 96%) were slightly higher than the sensitivity and specificity of impedance plethysmography (94% and 90%), using contrast venography as the "gold" standard.4 The specificity of both test declines considerably when used in the evaluation of asymptomatic patients (76% for ultrasonography, 29% for impedance plethysmography). While the sensitivity is increased by the use of serial tests if the initial test is negative, the vast majority of positive ultrasound tests (93% in one study) occur on the first test of the series.5
The most common signs and symptoms of pulmonary embolism are dyspnea, tachypnea, tachycardia, low-grade fever, pleuritic pain, and syncope. As with deep venous thrombosis, signs and symptoms are inadequate for diagnosis of PE. The ventilation-perfusion scan has been established as the appropriate initial diagnostic test for pulmonary embolism.6 The results of lung scans may be characterized as high,- intermediate,- and low-probability patterns and near-normal/normal for pulmonary embolus.7 A normal perfusion scan has a low chance of being associated with a pulmonary embolus (4% in the Prospective Investigation of Pulmonary Embolism Diagnosis [PIOPED] study).3,7 A high-probability scan demonstrating segmental or greater perfusion defects with normal ventilation is associated with pulmonary embolism in about 90% of cases and can be used to clinically establish a very high likelihood of pulmonary embolus.7 Current diagnostic strategies used in the face of intermediate- and low-probability scans include noninvasive testing for deep venous thrombosis, with treatment decision and the advisability of pulmonary arteriogram based on the presence or absence or DVT.6 The noninvasive diagnosis of DVT confers great benefit in this situation in that the need for pulmonary arteriogram is precluded, since treatment of pulmonary embolus and of DVT with anticoagulants is identical. Recently published data suggest that pulmonary arteriography should be performed if testing for DVT is negative, because more than 10% of these patients demonstrated a pulmonary embolism when arteriography was performed.5
For both DVT and PE, the level of clinical suspicion of the disease affects the predictive value of screening tests. In a study population of patients referred for evaluation of DVT, the prevalence of venography-proved DVT were 5%, 25%, and 80% in subgroups assessed by clinicians to have low,- intermediate,- and high-probability of disease.8 Using these values, Wheeler demonstrated that the positive predictive value of a diagnostic test with high (90%) sensitivity and high specificity (90%) was 97% for a patient with high clinical probability of DVT and only 32% for a patient with low clinical probability of DVT.4 Among participants in the PIOPED study with a scan showing a high probability of pulmonary embolism, the positive predicative value of the scan for arteriography-proved PE was 96% for those considered to have a high clinical probability of PE and 56% for those considered to have a low clinical probability of PE.7