Ultrasound in the diagnostic work-up of pulmonary embolism

Pulmonary embolism is a common cardiovascular condition which may present with a wide range of unspecific symptoms, such as dyspnea, chest pain, or tachycardia, making the diagnosis challenging. To avoid overuse of radiation diagnostics such as CT pulmonary angiography or ventilation/perfusion scans, international societal guidelines recommend that when suspicion of pulmonary embolism is raised, selection of patients for final diagnostics should be based on clinical prediction scores and measurement of D-dimer ^4–8.

The clinical utility of D-dimer is however complicated by the specificity which diminishes from≈ 67% in patients around 40 years of age to ≈ 10% at the age of 80 or in the presence of conditionssuch as cancer, infection, or inflammatory disease, resulting in a significant proportion of falsepositives^4,9. Consequently, the prevalence of pulmonary embolism in diagnostic imaging continuesto decrease globally and has recently been described as 20% in the European setting^10–12. This PhDthesis explores the utility of ultrasound as an aid in improving selection of patients for radiationimaging. 

We initially conducted a systematic review and meta-analysis on the diagnostic accuracy ofultrasound in suspected pulmonary embolism. The 70 included studies assessed either the lungs,heart, deep veins of the legs, or any combination thereof, reporting on subpleural infarctions, rightventricular strain, or deep venous thrombi. Meta-analysis revealed that some ultrasound signs arecharacterized by high sensitivities or specificities.

Based on our findings, we constructed a bespoke ultrasound protocol with the aim of allocatingpatients with suspected pulmonary embolism into one of three groups: 1) pulmonary embolismdismissed, 2) confirmed or, 3) need for radiation imaging. Our descriptive findings suggested that acombination of a low pre-test probability and ultrasound devoid of signs of pulmonary embolism hada high negative predictive value, while presence of the most specific ultrasound signs, such as apresent deep venous thrombus or McConnell’s sign is highly predictive of pulmonary embolism,regardless of pretest probability. 

To assess if this approach reduces referral to radiation diagnostics in a real-life setting, we concludedthe thesis by conducting a multicenter trial randomizing 150 patients to either standard diagnosticapproach or investigation with our bespoke multiorgan approach. We found a 45.2% (95% CI: 34.3to 56.6) reduction in referral for radiation diagnostics in the 73 patients randomized to the ultrasoundgroup. Of the 30 patients who did not receive anticoagulation, two (6.7%, 95% CI: 1.9 to 21.3) werediagnosed with pulmonary embolism during the three-month follow-up period.