Purpose: Large medical record databases facilitate epidemiology research in fracture. However, the validity of fracture in the databases is needed to ensure the reliability of data. We aimed to assess the validity of International Classification of Diseases, 9th Revision (ICD-9) code algorithms for identifying major osteoporotic fracture in the Clinical Data Analysis and Reporting System (CDARS) in Hong Kong.
Methods: The CDARS is a database developed by the Hong Kong Hospital Authority for research purpose. We used ICD-9 code algorithm for identifying major osteoporotic fracture, including vertebral fracture, humerus fracture, forearm/wrist fracture, and hip fracture, in CDARS in 2005-2016. As high positive predictive value (PPV) is critically important in epidemiology research, we sought to determine the PPV of fracture diagnostic code in terms of ICD-9 relative to the radiography imaging and clinical notes. A total of 380 major osteoporotic fracture cases (vertebral fracture: 101 cases; humerus fracture: 81 cases; forearm/wrist fracture: 94 cases; and hip fracture: 104 cases) were randomly selected and validated.
Results: In 380 fracture cases, the overall PPV was 96.8%. In subgroup analysis, PPV of 100% was observed for hip, humerus, and forearm/wrist fractures, whereas PPV of 86% was observed for vertebral fracture.
Conclusions: The use of ICD-9 code algorithm to identify major osteoporotic fracture in CDARS is a valid tool with a very high PPV. However, cautious interpretation is required when the study focuses on incident vertebral fracture. Copyright © 2017 John Wiley & Sons, Ltd.
Keywords: ICD-9; fracture; osteoporosis; pharmacoepidemiology; positive predictive value.
Copyright © 2017 John Wiley & Sons, Ltd.