Accuracy of plain abdominal radiographs in the detection of retained surgical needles in the peritoneal cavity

OBJECTIVE: To determine the accuracy of plain abdominal radiographs in the detection of retained surgical needles of varying size in the peritoneal cavity. SUMMARY BACKGROUND DATA: Accidental retention of surgical foreign bodies in the peritoneal cavity is estimated to occur once in every 1000 to 1500 abdominal operations and early prevention and identification of retained foreign bodies is increasingly important because of mounting public awareness. Most of the existing literature on the imaging detection of surgical foreign bodies has focused on retained sponges, even though retained needles may account for up to 50% of such objects and the true accuracy of plain abdominal radiographs in the detection of retained needles is not well established. METHODS: Eight plain radiographs were obtained of a 41 kg pig cadaver after placement of a total of 39 surgical needles of varying size (4-77 mm in length) in a randomized selection of the 9 segments of the peritoneal cavity. Five radiologists independently reviewed the radiographs and indicated the location of all suspected retained needles. Analyses were performed using the known site and size of placed needles as the standard of reference. RESULTS: In total for all readers, 195 needles were detectable in 360 abdominal segments. The overall mean accuracy, sensitivity, and specificity for plain radiographs in the detection of retained surgical needles were 74% (267 of 360), 69% (135 of 195), and 80% (132 of 165), respectively. Sensitivity for needles 25 mm or more in length was significantly (P < 0.0001) higher than that for needles of 11 to 24 mm or 10 mm or less, with respective values of 99% (69 of 70), 84% (46 of 55), and 29% (20 of 70). Readers demonstrated moderate interobserver agreement, with a multireader κ value of 0.60. CONCLUSIONS: Abdominal radiographs have high sensitivity and interobserver agreement in the detection of retained surgical needles over 10 mm in length, but smaller needles are detected with significantly lower sensitivity and the utility of plain abdominal radiographs in this setting is more debatable.