| Introduction. The 'real-world' effectiveness of osteoporosis prescription medications for the prevention and treatment of osteoporosis cannot be determined from randomized clinical trials. Using data from administrative health databases, this thesis presents a retrospective observational study methodology to specifically measure the risk of osteoporosis-related fractures in women treated outside the clinical trial arena (i.e. in the 'real-world'). However, several factors related to the receipt of a prescription therapy prevent the direct comparison of outcomes between treated and untreated individuals. This dissertation attempts to identify these factors and ultimately apply this information to an overall 'drug effectiveness' model.; Methods. Using population-based administrative health databases, a cohort of women over 50 years of age was identified. Pharmaceutical claims data from 1996 through 2002 were analyzed to identify new users of osteoporosis prescription medications (OSRx). Three separate Cox proportional hazards regression analyses were conducted to identify the factors associated with (a) the likelihood of initiating an OSRx; (b) the likelihood of receiving a bone mineral density (BMD) assessment; and (c) the likelihood of discontinuing treatment. Given the factors identified in each of these analyses, a propensity score was developed to stratify women according to their likelihood to initiate an OSRx, and a final stratified Cox proportional hazards model was developed to estimate the overall effect of OSRx treatment on fracture risk.; Results. A cohort of 112, 463 women were identified of which 12.5% were dispensed at least one OSRx between April 1st , 1998 and March 31st, 2002. Factors associated with the likelihood of treatment initiation included: a prior BMD assessment, age, prior osteoporosis-fracture status, income level, and region of residence. Given the strong predictive effect of a BMD assessment on the likelihood of treatment initiation, it was necessary to explore the factors associated with receiving this test. Above all other variables tested, the BMD-testing rate of a woman's physician most strongly predicted her likelihood of receiving a BMD test. (Abstract shortened by UMI.)... |
