Intra- and interobserver reliability for predicting hip preservation versus hip arthroplasty utilizing plain radiographs with comparison of surgeon specialization

Surgeon subspecialty training and practice landscape are formative in diagnostic evaluation and treatment recommendations. Varying recommendations can have substantial impact on patients' care pathways and outcomes. We investigated intra- and interobserver reliability of treatment predictions for total hip arthroplasty (THA) between surgeons performing arthroplasty and/or arthroscopic hip preservation surgery. Anterior-posterior (AP) hip radiographs cropped to include the lateral sourcil, medial sourcil and foveal region of 53 patients with Tonnis Grade 0-3 were evaluated by five surgeons (two performing arthroplasty, two performing arthroscopic hip preservation and one performing both interventions). Surgeons predicted THA versus no THA as the treatment for each image. Predictions were repeated three times with image order randomized, and intra- and interobserver reliability were calculated. Surgeons were blinded to patient characteristics and clinical information. Interobserver reliability was 0.452 whereas intraobserver reliability ranged from 0.270 to 0.690. Arthroscopic hip preservation surgeons were more likely to predict THA (36.9%) than arthroplasty surgeons (32.7%), P = 0.041. Intra- and interobserver reliabilities of surgeons predicting THA versus no THA based on an AP hip radiograph were average at best. Arthroscopic hip preservation surgeons were more likely to predict THA than arthroplasty surgeons. Subjective surgeon interpretation can lead to variability in recommendations to patients; potentially complicating care pathways.