SUMMARY: Hospital diagnosis codes are useful for assessing hip fracture rates in large populations. However, these codes do not reliably differentiate hip fractures that occur in the subtrochanteric region. Identification of subtrochanteric fractures requires review of radiographic images to distinguish these fractures from the more commonly occurring trochanteric fractures. PURPOSE: This study examines the accuracy of coded hospital diagnoses for hip fracture compared to fracture site verification based on operative and radiologic data. The variability in subtrochanteric fracture assignment was also examined using different anatomic criteria. METHODS: This retrospective study includes female members of Kaiser Permanente Northern California age 60 years and older with nontraumatic hip fracture during 2007-2008. Anatomic site was verified by operative and radiologic records, including radiographic image review for fractures occurring in the subtrochanteric region. Two different criteria were compared for subtrochanteric fracture. RESULTS: We identified 2,824 women with incident hip fracture during the 2-year period. The average age was 82.9 +/- 8.2 years and 15% were non-White. International Classification of Diseases, Ninth Revision (ICD-9) coding was accurate for femoral neck and trochanteric fractures (>90% confirmed by operative/radiologic reports), compared to only 26% for subtrochanteric fractures using the Orthopedic Trauma Association (OTA) criteria for subtrochanteric fracture. Using OTA classification, 1.3% of hip fractures were assigned as subtrochanteric compared to 4.2% when the criteria were broadened to include the lesser trochanter. Both femoral neck and pertrochanteric fracture rates increased exponentially with age, while age-related rates in subtrochanteric fracture differed by diagnostic classification method; the broader criteria including the lesser trochanter produced age-related trends that mirrored femoral neck and pertrochanteric fractures. CONCLUSION: Unlike femoral neck and pertrochanteric fractures, epidemiologic studies of subtrochanteric fractures cannot rely on ICD-9 codes alone. Review of radiologic images using OTA criteria is required for identification of subtrochanteric fractures occurring below the lesser trochanter.