Background: Evidence suggests that prolonged bisphosphonate (BP) treatment predisposes to atypical fractures (AF), but the etiology has yet to be determined. Addressing causality begins with case identification, which requires radiological adjudication. However, many trials based their case findings on coded diagnoses. 

Objectives: To investigate the feasibility of case findings by the coding system and the reproducibility of radiological evaluations in two hospitals in Israel, and to compare BP exposure of AF patients to a control group with typical (intertrochanteric of femoral neck) fractures. 

Methods: Diagnostic databases from 2007–2010 were reviewed and admission X-rays of patients were examined in two steps by two radiologists. Fractures were classified as atypical or not atypical according to published criteria. A 2:1 control group was created. Ambulatory drug acquisition was reviewed. 

Results: Of the 198 patients who fulfilled the search criteria, 38 were classified by initial radiological opinion as AF. Subsequent radiological opinion judged 16 as not atypical. Of the AF patients, 80% were exposed to BP. Of those, 81% continued to receive BP treatment for 2.4 years after AF. Only one AF patient was discharged with suspected AF diagnosis. In the control group, 27% were exposed to BP prior to fracture (P < 0.001). 

Conclusions: Thorough radiological revision is mandatory for proper classification of AF, and even when performed there is significant inconsistency in interpretation. Conclusions drawn from trials based solely on coded diagnoses lead to significant bias. BP exposure was significantly higher in the AF group. Caregiver unawareness of AF leads to improper management.