Appraising the Fracture Risk Assessment Tool (FRAX) in persons with rheumatoid arthritis

Persons with rheumatoid arthritis (RA) have up to two times higher risk of fracture than age and sex-matched comparators. These fragility fractures lead to significant morbidity and excess mortality in persons with RA. Most fragility fractures occur in persons with bone mineral density above the "osteoporotic" cutoffs (i.e. osteopenia), a group where fracture risk assessment tools are paramount to guide therapeutic decisions. In clinical practice, FRAXTM is the most widely used fracture risk prediction tool in persons with RA. FRAX computes 10-year risk of major osteoporotic fracture (MOF; hip, clinical spine, forearm, or humerus fracture) and hip fracture using clinical fracture risk factors. Recent studies suggest FRAX without BMD overestimates MOF and hip fracture risk in persons with RA, while FRAX with BMD may underestimate MOF risk in middle- aged persons with RA. Existing validations studies of FRAX use data from predominately White cohorts only. Despite its widespread use, the predictive ability of FRAX-U.S. in a more contemporary and diverse U.S. RA population is unknown. Our overarching objective is to optimize fracture risk prediction in persons with RA. To achieve this objective, we aim to evaluate the predictive validity of FRAX-U.S. for 10-year risk of MOF and hip fracture in persons with RA and compare predictive performance with the general population using 1) a cohort of women (75%) and men with RA age ≥ 65 years and 1:1 matched controls who are Medicare beneficiaries enrolled in Part D pharmacy benefits and 2) a cohort of women and men Veterans with RA age ≥ 40 years and 1:3 matched controls in the Veterans Health Administration. Subgroup analyses will assess FRAX-U.S. performance by sex, race/ethnicity, and age categories. We further aim to determine the impact of additional potential fracture risk factors related to or mediated by RA on the predictive performance of FRAX-U.S. in persons with RA. Completion of this research proposal is only one component of the thoughtfully developed mentored career development plan which provides the fundamental first step towards achieving my short-term goals to (1) obtain detailed knowledge and expertise in curating, analyzing, and interpreting Medicare data for epidemiologic studies and (2) attain new skills in data science, informatics, natural language processing, and machine learning. This application is a vital first step toward my long-term goal to become a physician scientist with a focus on clinical-based epidemiologic research to improve the health of persons with RA and other rheumatic diseases. The training activities I will undergo include a mixture of a strong mentorship framework, experience working with a new data source to me (Medicare), a well-circumscribed curriculum organized towards obtaining a Master of Science in Analytics, a national rheumatology research networking and career development program, and informal seminars to bring me together with early investigator peers across public health disciplines at AU and a Center for Healthy Aging with weekly meetings that connects basic science, translational and clinical researchers at all career stages for intellectual discourse and productive collaboration.