Physical activity and osteoporotic fracture risk in older men: An application of causal inference methods with observational data

Osteoporotic fractures are a major public health problem for the elderly population and are associated with substantial morbidity, mortality, and cost. As the population ages, effective fracture prevention strategies are urgently needed. Physical activity may reduce fracture risk by improving bone strength and reducing falls. The aims of this dissertation were to (1) describe physical activity patterns among older men, and (2) estimate the marginal, population-level effects of total amount of physical activity on hip and non-vertebral fracture risk among older men.;Participants were enrolled in the Osteoporotic Fractures in Men study. Physical activity was measured at baseline with the Physical Activity Scale for the Elderly. Incident non-vertebral fractures were confirmed by centralized physician review of radiology reports over 6.5 years of follow-up. Data on potential confounding factors were obtained. To facilitate causal inference, marginal effects of physical activity on hip and non-vertebral fracture risk were estimated with marginal structural models using inverse probability of treatment and censoring weighted, G-computation, and targeted maximum likelihood estimation methods.;Participants were exceptionally active: 80% met the national recommendation for moderate-intensity aerobic activity; and 42% met the recommendation for muscle strength and endurance activities. Leisure-time activities accounted for a smaller proportion (25%) of overall activity than household activities (64%).;Compared to men with low physical activity, crude hip fracture risk was lower among men with high physical activity, and crude non-vertebral fracture risk was lower among men with moderate physical activity. However, causal inference analyses, which controlled for confounding and potentially informative censoring, demonstrated that neither hip nor non-vertebral fracture risk would be reduced if all men participated in moderate or high physical activity relative to low physical activity. Osteoporosis status did not appear to modify these results. Population-level uptake of high levels of physical activity may elevate non-vertebral fracture risk.;In conclusion, typical patterns of physical activity do not protect against hip or non-vertebral fracture in older men. Future research may identify more specific patterns of physical activity that reduce fracture risk in this population.