Interobserver variation in the interpretation of abdominal radiographs

A total of 140 sets of abdominal radiographs were reviewed independently by four qualified diagnostic radiologists. The degree of interobserver agreement was determined by calculating kappa values for 19 commonly used radiographic signs and diagnoses. There was fair to excellent interobserver agreement for 11 signs and diagnoses and poor agreement for the remaining eight. The signs and diagnoses for which agreement is poor cannot be considered reliable and include particularly large bowel obstruction and nonspecific gas pattern.