Bedside ultrasound for the detection of soft tissue foreign bodies: a cadaveric study

The objective of this study was to evaluate the sensitivity and specificity of bedside ultrasound, as performed by emergency physicians with typical equipment, in detecting small, soft tissue foreign bodies, using a cadaveric model. This was a prospective study, using 6 unembalmed human cadavers and 6 ultrasound-credentialed, emergency medicine residency-trained physicians as sonographers. Incisions were made in 150 total sites of the extremities and each site was randomly assigned one of five groups: wood, metal, plastic, glass, or no foreign body. All foreign bodies were 2.5 mm3 in total volume or less, no longer than 5 mm in any dimension, and inserted to a depth of up to 3 cm. Ultrasound was performed with a SonoSite TITAN (SonoSite, Inc., Bothell, WA) ultrasound system using a L38/10-5 broadband linear array transducer. Sonographers were blinded to the number, type, and location of foreign bodies. A total of 900 ultrasound examinations were recorded. Overall sensitivity of ultrasound for foreign body detection was 52.6% (95% confidence interval [CI] 48.9%-56.2%), and overall specificity was 47.2% (95% CI 39.9%-54.5%). Positive predictive value was 79.9% (95% CI 76.3%-83.5%), and negative predictive value was 20.0% (95% CI 16.2%-23.7%). Sensitivity for individual sonographers ranged from 40.8% to 72.3% (average 52.6% +/- 13.3%), and specificity ranged from 30% to 66.7% (average 47.2% +/- 15.1%). Inter-observer reliability was poor. In our model, bedside ultrasound performed by emergency physicians was neither sensitive nor specific for the presence of small soft tissue foreign bodies.