ABSTRACT
<p><b>Objective</b>: Monitoring the incidence of atypical femoral fractures (AFFs) using medical claim databases is useful to assess the safety of long-term bisphosphonate exposure. Therefore, we aimed to validate the relationship between clinically-defined suspected AFFs and the candidate patients obtained from claims data at three hospitals in Japan.</p><p><b>Design</b>: A cross-sectional study involving three hospitals that perform bone fracture surgery and from which electronic medical record databases of diagnoses and procedures are available.</p><p><b>Methods</b>: Candidate patients were at the medical databases using two International Classification of Diseases, 10th Edition (ICD-10) codes (subtrochanteric fracture and fracture of shaft of femur) in the claims databases. These potential cases by claim-based definition were validated using clinically-confirmed information such as, the patient operation records, the discharge records, or radiographic imaging findings as suspected AFFs.</p><p><b>Results</b>: Among fracture cases in the hospitals, and 9 cases with subtrochanteric fracture and 23 cases with femoral shaft fracture were identified based on the ICD-10 codes in the claims databases. Clinically confirmed subtrochanteric fracture had a sensitivity of 81.8% (95% CI: 48.2-97.7%), and a specificity of 100.0% (95% CI: 99.9-100.0%). For femoral shaft fracture, the sensitivity was 82.1% (95% CI: 63.1-93.9%), and the specificity was 100.0% (95% CI: 99.9-100.0%). In subgroup analyses, the sensitivities in patients over the age of 50 years with a single fracture site and with osteoporosis were relatively higher than in other subgroups.</p><p><b>Conclusion</b>: The claims-based definitions of suspected AFFs are accurate, indicating the value of pharmacoepidemiological studies using the National Receipt Database.</p>