RELIABILITY IN RADIOGRAPHIC REVIEW OF TIBIAL SPINE FRACTURES IN A TIBIAL SPINE RESEARCH INTEREST GROUP

BACKGROUND: Variability that exists amongst the radiographic measurement parameters associated with tibial spine fractures may have direct consequences when comparing, reporting, or treating these injuries. In developing data collection of tibial spine fractures amongst multiple centers, it is important to establish reliability in radiographic parameters. Therefore, we designed a study to validate the classification and a proposed cohort of measurements of tibial spine fractures amongst multiple institutions to assist with standardizing fracture classification and treatment decisions. METHODS: Radiographic assessment of de-identified acute tibial spine fractures was performed by members of the Pediatric Research in Sports Medicine (PRISM) Tibial Spine Research Interest Group. A descriptive Powerpoint presentation was provided to each reviewer demonstrating specific measurements and classification prior to review. Reviewers were also asked to provide treatment recommendations. DICOM files were provided to the surgeon through a web-based shared drive and reviewers were required to use the same imaging software. There were 40 patients included, determined through power analysis performed based on previous reliability studies and the number of participants. Assuming the intraclass correlation coefficient (ICC) will be .85 and 95% confidence interval to be 0.2, the sample size of 40 will achieve the desired 95% confidence. Data will be reviewed using both kappa and ICC reliability measures due to both categorical and continuous data points. RESULTS: A majority of radiographic measures demonstrated moderate ICC including posterior-proximal displacement (0.378), length and height of tibial spine fracture (0.466 and 0.535, respectively), and superior displacement of medial fragment (0.420). Good ICC was seen with superior displacement of the anterior tibial spine fragment (0.734). Poor correlation was seen with the measurements for anterior displacement, posterior sagittal displacement, and roof inclination angle. Classifying tibial spine fractures according to the historical Meyer & McKeever Classification demonstrated fair agreement (kappa = 0.347). 18 of 40 (45%) fracture patterns were classified by reviewers in three or more different classifications types while only 1 fracture pattern (Type 1) was agreed upon by all reviewers. A majority of reviewers recommended arthroscopic treatment with suture for more fracture patterns. However, there was fair agreement w