Intraobserver and interobserver evaluation in the radiological diagnosis of bisphosphonate-related fractures

Abstract

ObjetiveEvaluate the utility of radiography for diagnosing bisphosphonate-related atypical fractures. MethodsPatients with a subtrochanteric and diaphyseal femur fracture, admitted to the institution from 2006 to 2011, have been retrospectively reviewed. Exclusion criteria included neck, intertrochanteric and supracondylar femoral fractures, pathological fractures and paediatric patients. All clinical data of 167 patients were reviewed from case records of which 29 had a bisphosphonate-related atypical femoral fractures.Out of 169 fractures classified (including 2 bilateral patients), 109 were diaphyseal and 60 subtrochanteric.Seven MD orthopaedics divided in 3 groups reviewed all radiographs of the 169 fractures retrospectively, independently and twice. The three groups readers were different expertise levels (group 1: 3 residents in the third year; group 2: one fellowship in knee and the other one in shoulder; group 3: one expertise in hip prothesis and another in foot and ankle).The following information has been included in the readers’ data collection sheets: focal lateral cortical thickening, transverse orientation of fracture and medial spike.The three fracture characteristics were assessed as present or absent and were recorded as: YES for three criteria present,  DON´T KNOW  for two criteria present and NO for 0 criteria present. The intraobserver agreement was calculated with  kappa coefficients  while the interobserver agreement was calculated with the G-string program. The statistical analysis applied was YES for data with 3 criteria. ResultThe intraobserver agreement was kappa = 0.2 to 0.5. The Interobserver agreement was 54 % average.  ConclusionRadiographs alone are not reliable enough for distinguishing complete femoral fractures related to bisphosphonate.