The overall goals of this competing continuation application are to: 1) obtain precise estimate of the lifetime risk of fractures for women, and 2) identify geographical differences in fracture risks in the U.S. for all race sex groups and relate them to risk factors. The long-term goals are to develop more rational strategies for prevention of osteoporosis in women and to identify new risk factors for osteoporosis.
The specific aims are to develop statistical methods and apply them to existing datasets to achieve these goals. The first goal will be achieved through four specific aims: 1) estimating the correlation of baseline radius bone mass measurement with subsequent measurement at other sites, 2) predicting non-spine fracture from the presence of early spine fractures and early rates of bone loss, 3) estimating the correlation of an individual's radius bone mass measurement over time, and 4) formulating a model to predict lifetime fracture risks based on the first three findings. These analyses will be based on data collected from a longitudinal study of 521 healthy women recruited between 1971 and 1983. Multiple measurements of radius bone mass and fracture occurrences have been recorded on all subjects, and bone mass at all sites will be made on the 150 active subjects within the next year. Reports from spine x-rays on the first 300 recruits are also available and the original films saved on 120 subjects will be re-read uniformly by a radiologist. Standard statistical procedures will be used or new ones will be developed and applied to the data for each of the 4 aims. The second goal will be addressed through 2 specific aims: 1) to improve the estimation of fracture incidence at the county level in the U.S. using Medicare claims data, and 2) identify potential risk factors at the county level which are associated with incidence of hip fracture. Fractures will be identified from Medicare and VA hospital claims databases. The rate estimates based on denominators from the U.S. censuses and Medicare enrollee files will be compared. Verification of hip fractures will be based on 2 algorithms developed independently by two research groups. Empirical Bayes estimates of the county-level hip fracture incidence rates will be calculated.The stabilized rate estimates will be correlated with potential risk factors from other public use datasets such as the Area Resource File and NHANES III when the data become available.
Showing the most recent 10 out of 22 publications
