Surgeon variability in the treatment of pelvic ring injuries

Pelvic ring injuries are associated with the potential for long-term disability and high mortality rates. No well-established, definitive treatment algorithms have been presented in the literature. We evaluated agreement among surgeons in selecting treatment of pelvic ring injuries and investigated the relationship between type of injury and treatment plan. We conducted a prospective agreement analysis of pelvic ring injury images at a level I regional trauma center. Eighty-nine isolated pelvic ring disruptions were selected; they were the only injuries present on the images and were thought to comprise a variety of pelvic fracture types from a database of 1600 pelvic fractures. Sets of injury images were randomly ordered and distributed to 5 orthopedic trauma surgeons blinded to patient name, attending surgeon, date of injury, and eventual treatment. Surgeons reviewed images, independently selected preferred treatments, and classified injuries with the Young-Burgess and Tile systems. Eight weeks later, images were again randomly ordered and distributed to the same 5 orthopedic surgeons, who again independently selected preferred treatments and classified injuries. Kappa analyses of agreement among surgeons and within each surgeon over time were conducted. Analysis among surgeons' treatment plans based on radiographic assessment alone revealed a mean kappa value of 0.47, indicating moderate level of agreement. Analysis of data collected 8 weeks later yielded a mean kappa value of 0.56, representing moderate agreement within each surgeon over time. Young-Burgess and Tile classifications yielded consistent treatment plans for certain fracture types and varied plans for other types. Our results question the usefulness of the 2 classification systems for predicting treatment decisions.