Accuracy of compression ultrasound in screening for deep venous thrombosis in acutely ill medical patients.

The value of vein ultrasonography for diagnosis of symptomatic deep vein thrombosis (DVT) is widely accepted. We are unaware of published data comparing ultrasonography with the "gold standard" of venography for DVT diagnosis in asymptomatic persons in the patient group of acutely ill medical patients. It was the objective of this study to evaluate sensitivity and specificity of compression ultrasound (CUS) examinations in the diagnosis of proximal and distal DVT in acutely ill medical patients [with congestive heart failure (NYHA class III and IV), exacerbations of respiratory disease, infectious disease, and inflammatory diseases] considered to be at moderate risk of venous thromboembolism (VTE). CUS examination was performed prior to ascending venography on day 6-15 of the hospital stay. Both investigations were done on the same day, each interpreted without knowledge of the other's result. Proximal and calf veins were separately evaluated. Technically satisfactory venography was obtained in 160 patients. In 12 of 160 patients (7.5%, 95% CI=[4.0%-12.7%]), venography confirmed the presence of DVT, all of which was asymptomatic. Proximal DVT was detected in five patients (3.1%, 95% CI=[1.0%-7.1%]) and distal DVT in seven patients (4.4%, 95% CI=[1.8%-8.8%]). CUS of proximal veins was technically satisfactory in all 160 patients and CUS of distal veins in 150 patients. In three of five patients with venographically proven proximal DVT, the diagnosis was confirmed by CUS (sensitivity 60%, 95%CI=[23%-88%]). In one patient, the CUS was false positive (specificity 99.4%, 95%CI=[96%-99%]). Positive and negative predictive values (PPV and NPV) of CUS in the diagnosis of proximal DVT were 75% (95%CI=[30%-95%]) and 98% (95% CI=[95%-99%]), respectively. In two of seven patients with venographically proven calf DVT, the diagnosis was confirmed by CUS (sensitivity 28.6%, 95%CI=[8%-64%]) and in two patients, CUS was false positive (specificity 98.6, 95%CI=[95%-99%]). PPV and NPV of CUS in diagnosis of distal DVT were 50% (95%CI=[15-85%]) and 96% (95% CI=[92%-98%]), respectively. In conclusion, CUS underestimates the incidence of proximal and distal DVT compared to contrast venography in acutely ill medical patients without thrombosis symptoms.