To determine the sensitivity and specificity of portable fluoroscopy for foreign-body detection using an ex-vivo experimental model. A prospective, randomized masked investigation was performed to characterize foreign-body detection in cubes of beef. Three emergency physicians (EPs) independently used a portable fluoroscope to determine the presence or absence of various foreign bodies in ten cubes of beef (approximately 6 cm on each side). Six different groups of foreign bodies were examined: glass, wood, graphite (pencil lead), metal (needle), plastic, and gravel. An independent observer placed the six objects randomly into each set of ten beef cubes. One hundred observations were made by each physician using sets of ten beef cubes at a time. Fluoroscopy detected 117 of 180 foreign bodies (sensitivity 65%; 95% CI 58-72%), including all cubes of beef embedded with gravel, metal, and glass. Graphite was detected 27 of 30 times (sensitivity 90%; 95% CI 74-98%). None of the plastic or wooden foreign bodies was identified despite magnification techniques and real-time imaging (sensitivity 0%; 95% CI0-12%, each). Of the 120 cubes of meat with no foreign body, two false-positives were recorded (specificity 98%; 95% CI 94-100%). The overall positive predictive value and negative predictive value of fluoroscopy were 98% and 65%, with 95% CIs of 94-100% and 58-72%, respectively. Interobserver agreement between the three EPs was considered excellent (kappa = 0.75). Bedside fluoroscopy lacks sufficient sensitivity to rule out many foreign bodies in the ED. Its use should be limited to suspected gravel, glass, pencil lead, and metallic objects, which are known to be radiopaque.
Read full abstract