Aims and objectives
Prevalence of pulmonary embolism (PE) in patients referred to CT pulmonary angiography continues to decline. As such, improving selection of patients for diagnostic imaging while maintaining an acceptable failure rate is warranted. The aim of this study was to assess reduction in referral to diagnostic imaging and failure rate by using a bespoke ultrasound protocol.

Methods
In a multicenter trial, adult patients with suspected PE and a Wells score of 0-6 and elevated age-adjusted D-dimer or Wells score >6 were randomized 1:1 to direct diagnostic imaging (controls) or focused lung, cardiac and deep venous ultrasound. Ultrasound could: 1) dismiss PE if no signs of PE and low clinical suspicion or an alternate diagnosis, 2) confirm PE in case of visible venous thrombus, ≥2 subpleural infarctions, McConnell’s, or D-sign, or 3) refer to diagnostic imaging if neither category was fulfilled or a patient with confirmed PE required admission. Primary endpoint was reduction in referral to diagnostic imaging. Secondary endpoint was PE diagnosis during 3-month follow-up (failure rate).

Results
Of 150 patients, 73 were randomized to ultrasound. In controls, 26 PEs were confirmed. In the ultrasound group, 40 patients underwent diagnostic imaging of whom 20 had PE, reducing diagnostic imaging by 45.2% (95% CI: 34.3 to 56.6). 3 further PEs were diagnosed by presence of a DVT. In the ultrasound group 30 patients (41%) had PE dismissed of whom 2 were diagnosed with PE during follow-up, corresponding to a failure rate of 6.7% (95% CI: 1.9 to 21.3). 

Conclusions
Ultrasound substantially reduced referral to diagnostic imaging albeit with a high failure rate.