Accuracy of nurse-performed compression ultrasonography in the diagnosis of proximal symptomatic deep vein thrombosis: a prospective cohort study

Background Compression ultrasonography (CUS) has been recognized as the diagnostic procedure of choice for the investigation of patients with suspected deep vein thrombosis (DVT); the aim of this study was to assess the diagnostic accuracy of nurse-performed CUS for symptomatic proximal DVT of the lower limb. Material and Methods We prospectively evaluated all consecutive outpatients referred for suspected DVT from January 2011 to December 2012. All patients underwent bilateral proximal lower limb CUS, first by trained nurses and then by physicians expert in vascular ultrasonography, with every group blinded with respect to each other. This test was repeated after 5-7days in all negative or unclear examinations. Interobserver agreement and accuracy of nurse-performed CUS were calculated, considering the physician's final diagnosis as the reference test. Results Six hundred ninety-seven patients were included in the study. DVT was diagnosed in 122 patients by expert ultrasound physicians with an overall prevalence of 17.5% (95% confidence interval [CI] 15.8%-20.6%). Nurse agreement with the physician in DVT diagnosis was excellent (Cohen's kappa 0.82, 95% CI 0.79-0.85). Nurse-performed CUS had a sensitivity of 84.4% (95% CI 81.7%-87.1%) and a specificity of 97.0% (95% CI 95.8%-98.3%) with a diagnostic accuracy of 94.8% (95% CI 93.2%-96.5%). Conclusion Our results suggest that nurse-performed CUS may be a potential useful alternative to physician performed CUS with a good accuracy. However, sensibility of nurse-performed CUS appeared suboptimal and future studies should incorporate in the evaluation of this technique other pretest tools that may increase its accuracy.