Management of proximal humeral fractures: surgeons don't agree.

HYPOTHESIS The management options for proximal humeral fractures have expanded in recent years. Patients with displaced, unstable proximal humeral fractures may have improved outcomes if managed operatively. We investigated the decision making of fellowship-trained orthopedic surgeons when presented with the same group of cases. We hypothesized that interobserver and intraobserver agreement for surgical management would be poor and independent of fellowship training. METHOD Eight fellowship-trained orthopedic surgeons (3 shoulder, 5 trauma) viewed the preoperative plain radiographs of patients with proximal humeral fractures. All surgeons viewed the same 38 radiographs in a blinded fashion. Surgeons chose from 1 of 6 management options. Interobserver variability was calculated by using the weighted kappa coefficient. Intraobserver variability was calculated by comparing each surgeon's survey results with the operation they originally performed. RESULTS Overall interobserver agreement on management was moderate (weighted kappa=0.41) and did not differ significantly between trauma surgeons and shoulder surgeons. Reducing the number of management choices increased agreement between all surgeons. Testing for intraobserver agreement showed that surgeons picked the same operation in the survey as in the actual clinical setting only 56% of the time. CONCLUSION Interobserver agreement was moderate overall and improved when the number of management choices was reduced. Intraobserver agreement was less frequent, however, raising the question about consistent decision making by a given surgeon. Although surgeons agree in the method of treatment only to a modest degree, it remains for further outcomes research to establish if the choice of treatment actually influences the clinical outcome.