Radiological Assessment of Atypical Femoral Fractures: A 10-Year Experience Across Two Centres.

Objectives: Atypical femoral fractures (AFFs) are stress or insufficiency fractures occurring in the femoral shaft associated with long-term osteoporosis treatment with antiresorptives. Expeditious diagnosis of AFF is important, as ongoing treatment with antiresorptives is then contraindicated. AFFs can occur bilaterally, either simultaneously or sequentially; prompt diagnosis is pivotal to ensuring appropriate surgical and medical management. The American Society for Bone and Mineral Research (ASBMR) has devised diagnostic criteria based on history and distinctive radiographic features that are qualitative and subjective. Recently, the Sydney AFF Score (SS) has been devised which is more quantitative and objective, based on the age of the patient and two specific radiographic measurements. This study aimed to assess the global performance, sensitivity and specificity of the SS in this cohort. Secondary aims were to assess interobserver agreement in the assessment of the ASBMR criteria and SS measurements, the accuracy of radiological reporting and the utility of ICD-10 codes for diagnosis of AFF. Design: Retrospective observational study of patients presenting with AFFs or 'typical' femoral fractures (TFFs) as diagnosed by an expert panel. Setting: Two teaching hospitals in Queensland, Australia. Primary outcomes measure(s): (1) diagnosis of AFF using the SS; (2) measuring sensitivity and specificity of each ASBMR criterion for AFF; (3) accuracy of radiology reporting; (4) accuracy of ICD-10 codes in identifying AFF. Methods: Adult patients aged 18 years or older presenting with low-energy mechanism, subtrochanteric fractures between January 2012 and February 2022 were screened from the medical records using ICD-10 codes. A total of 46 AFFs were identified following expert panel review, with 38 TFFs in a comparator group. The sensitivity and specificity of both the SS (cortical width ≥ 5 mm at lesser trochanter, femoral neck width ≤ 37 mm and age < 80 years) and each of the ASBMR criteria was assessed. The degree of interobserver variability was measured using the Kappa statistic. Results: Out of 869 fractures, 46 AFFs were identified following independent expert panel review by a 2/3 majority, and a comparator group of 38 female patients was selected from the remaining 72 TFFs. The major radiological ASBMR criteria demonstrating highest diagnostic performance were transverse orientation (91.4%, 89.1%), minimal comminution (99.2%, 63.9%) and periosteal reaction (89.4%, 96.4%). The ASBMR criteria showed good interobserver agreement (Kappa), transverse orientation (0.72), minimal comminution (0.64) and local periosteal reaction (0.72), whereas the agreement with the anthropometric indices using the SS was less robust: lateral cortex ≥ 5 mm (0.25) and femoral neck width ≤ 37 mm (0.66). A 'modified' SS was created by incorporating the 'transverse' criterion. Of the AFFs, 13.2% of radiology reports in medical records referenced atypical or drug-related fracture. Conclusions: Qualitative indices of the ASBMR outperformed the SS. Incorporating transverse fracture orientation into the SS and using more specific landmarks for measurements may yield superior diagnostic value and help to mitigate subjectivity in the assessment of radiographs. This study confirms that AFFs are under-reported in pelvic X-rays.