Statistical methods for vitamin D targets for functional outcomes in older adults



 DESCRIPTION (provided by applicant): The concentration of 25-hydroxyvitamin D [25(OH)D] needed for older adults to maintain optimal physical function is unknown. Institute of Medicine and Endocrine Society have publicly debated on 20 ng/mL or 30 ng/mL, respectively, as targets for skeletal health, but both groups agree that stronger evidence is needed to identify target concentrations for extra-skeletal outcomes, e.g., falls and physical function. The problem is significant: >25% of older adults have poor physical function, 50% have 25(OH)D concentrations between 20 and 30 ng/mL. Identifying 25(OH)D targets for older adults will have immediate impact by informing design and analysis of vitamin D supplementation trials. Namely, targets serve as inclusion/exclusion criteria or baseline subgroups and as secondary outcomes. Our solution is to empirically identify and validate sex-specific (possibly stratified by other factors; e.g., race, genetic variants) 25(OH)D targets in older adults (= 65 years) related to lowe risk of falls and better physical function using existing NIH-funded datasets. A critical challenge is that 25(OH)D cannot be randomized; therefore observational data are a crucial part of the solution. Conventional statistical methods are ill-equipped for this problem; thus, we will develop novel statistical methods by adapting and refining structural modeling and machine-learning methods, and apply them to our aims. We will carry out the aims using 4 prospective cohort studies comprising >12,500 older adults: Health, Aging, and Body Composition, Invecchiare in Chianti, Osteoporotic Fractures in Men, and Study of Osteoporotic Fractures. We will: 1) identify and validate 25(OH)D targets for falls reduction, mobility, activities of daily living, and instrumental activities of daily living (functional outcomes); and gait speed, muscle strength, and frailty (proximal conditions) (aim 1); 2) characterize the joint impact of 25(OH)D and serum calcium and phosphorus concentrations on functional outcomes/proximal conditions (aim 2); and 3) quantify mediation of 25(OH)D on functional outcomes/proximal conditions by endocrine and metabolic factors (PTH, serum alkaline phosphatase, aklotho, fibroblast growth factor 23) (aim 3). We will write code in SAS and R statistical languages and disseminate it to the scientific community for use in additional research.