Modeling fracture prediction in spinal cord injury and disease

DESCRIPTION (provided by applicant): Anticipated Impacts on Veteran's Healthcare: The purpose of this project is to create a working tool to estimate fracture risk in Veterans with SCI/D. This tool will help guide clinicians in decision making for which patients with SCI/D should be treated with pharmacological agents to prevent fracture. Project Background: More than 10% of all persons with a SCI/D receive care through the VA and care of Veterans with SCI/D is of central concern to the VA. We have identified that serious morbidities occur following osteoporotic fractures in Veterans with SCI/D. However, there is very little known concerning risk factors for osteoporotic fractures in SCI/D. A better understanding of risk factors for fractures in SCI/D would allow appropriate targeting of existing treatments for osteoporosis. Recently, serious side effects from pharmacological therapies for osteoporosis including osteonecrosis of the jaw and atypical femoral fractures have been reported. Thus, it has now become critically important to avoid overtreatment of patients at low risk for fracture. Tools which incorporate clinical factors and bone mineral density (BMD) measurements to estimate fracture risk have recently been developed for persons without SCI/D. However, since the pathophysiology of bone loss following SCI/D differs substantially from that of age related, postmenopausal, or even simple disuse osteoporosis, there is a need to develop a distinct model to predict fractures in SCI/D. As a first step towards developing a working model to predict fractures in SCI/D, it is critical to determine whether Dual Energy X-ray Absorptiometry (DXA)-derived BMD is indeed useful for predicting fractures. A second step in developing a working model to predict fractures in SCI/D is to determine which clinical risk factors can predict incident fractures. A final model that clearly and easily identifes which patients with SCI/D are at highest risk for fracture that can be used directly in the clinics would facilitate decision making for clinicians. Such a tool could avoid overtreatment and potential harm to those at low risk for fracture by treatments for osteoporosis, and confer the benefits of such treatments to those who are most likely to fracture. In this context, our working hypothesis is that clinical factors but not DXA-derived BMD will best predict fractures in persons with SCI/D. Our hypothesis will be examined in the following specific objectives. Specific Objective 1: Determine the utility of DXA to predict fractures in Veterans with SCI/D. We hypothesize that DXA-derived BMD will not be able to predict incident fractures in SCI/D. To test our hypothesis, using a retrospective cohort design, we will include Veterans in the SCD Registry in FY2002-2012 and determine whether DXA-derived BMD is able to predict incident fractures during the study period. Specific Objective 2: Determine the best model for non-axial fracture prediction in Veterans with SCI/D. We hypothesize that a set of clinical factors will predict incident fractures in SCI/D. To test our hypothesis, in a retrospective cohort design, we will include Veterans in the SCD Registry in FY2002-2012 and determine which clinical factors can predict risk for incident non-axial fractures through FY2012. Our final fracture prediction model will include BMD (only if BMD can predict fractures) and those clinical risk factors identified to predict incident fractures. The immediate objectives of this research are to determine whether DXA-derived BMD is a useful tool in predicting fractures in SCI/D and to develop a model to predict fractures in SCI/D. The long-term objectives are to use this model to guide treatment decisions for fracture prevention in persons with SCI/D. Project Methods: This project will utilize administrative datasets and chart review data. The resources at the VA including the VA Spinal Cord Disorders (SCD) Registry, the DSS Pharmacy and Radiology Systems, the Corporate Data Warehouse and the Austin National Patient Care Databases will provide the administrative data for this project.