Identification and topographic localization of metallic foreign bodies by metal detector

Exact localization of ingested metal objects is necessary to guide therapy. This study prospectively evaluates the accuracy of foreign body (FB) identification and localization by metal detector (MTD) in a systematic topographic fashion. Patients who presented after an alleged or witnessed metal FB ingestion were scanned with an MTD. In case of a positive signal, the location was recorded in a topographic diagram, and radiographs were obtained. The diagnostic accuracy of the MTD scan for FB identification and topographic localization was determined by χ 2 analysis, and concordance was calculated by the McNemar test and expressed as κ. A total of 70 MTD examinations were performed on 65 patients (age 6 months to 16 years); 5 patients were scanned twice on different days. The majority had swallowed coins and button batteries (n = 41). Of these, 29 items were correctly identified, and 11 of 12 were correctly ruled out (coins and button batteries: sensitivity, 100% [95% Confidence Interval 95% to 100%]; specificity, 91.7% [95% CI 76% to 100%], κ = 0.94). When all metallic objects were included, 41 of 46 were correctly identified, and 22 of 24 were correctly ruled out (sensitivity, 89.1% [95% CI 80% to 98%]; specificity, 91.7% [95% CI 81% to 100%], κ = 0.78). Five miscellaneous objects were not identified (sensitivity for items other than coins and button batteries 71% [95% CI 49% to 92%], κ = 0.56). Localization by MTD was correct in 30 of 41 identified objects (73%). The error rates of junior and senior pediatric surgery residents did not differ significantly ( P = .82). Ingested coins and button batteries can be safely and accurately found by metal detector. For these indications, the MTD is a radiation-free diagnostic alternative to conventional radiographs. Other items, however, cannot be ruled out reliably by MTD. In these cases, radiographic imaging is still indicated.