Project Summary Use of bisphosphonate (BP) medications for the prevention of fractures is declining, partially due to patient and provider fears about the occurrence of atypical femur fractures (AFF), which are clearly associated with long- term bisphosphonate use. Many individuals are choosing to not use these medications at all rather than risk having this rare outcome. This decision may leave individuals at high risk of morbidity and mortality. Balancing the risk of typical osteoporosis-related fractures, such as hip or vertebral fractures, with the risk of the much rarer AFF, is an important component of decision-making by physicians and patients about medication use. Further, an understanding of which subgroups of women may be at greater or lesser risk of AFF will be useful when making clinical decisions about initiation of medication, duration of treatment, and use of drug holidays. The proposed study will address these issues by combining individual-level data from three large, population-based cohort studies with radiographically verified AFFs, comprehensive longitudinal medication exposure, data harmonized definitions of other covariables, and centrally coordinated statistical programming. We will be focusing on two Specific Aims. In Aim 1, we will examine the risks of long-term use of BP for the prevention of AFF by determining the independent effects of BP treatment and drug holidays on AFF risk, including the potential interplay between pre-holiday duration of treatment and duration of holiday. In novel exploratory analyses, we will determine the effects of re-initiation of BPs after a drug holiday. We will also evaluate the relationships between AFF risk and the use of other fracture prevention medications (e.g., intravenous BPs, denosumab and SERMs), other potential risk factors (e.g., physical activity, bone mineral density (BMD), race), and comorbidities. In Aim 2, we will use the pooled data from our 3 cohorts to develop and validate predictive models incorporating patterns of long-term BP treatment, drug holidays and clinical risk factors, to comprehensively model the expected fracture protection and potential harms for individual patients and evaluate the group-level balance between AFF risk and osteoporosis-related fracture prevention. These models will allow clinicians to quantify individualized risk, balancing the benefits and harms of bisphosphonate treatment accounting for other risk factors. Development of these predictive tools and addressing important gaps in the scientific evidence related to BP treatment, drug holidays, and re-initiation of BP or other anti-osteoporosis medications will have an immediate positive impact on clinical practice and patient care by encouraging and improving the optimal use of osteoporosis medications.