PROJECT SUMMARY Osteoporotic fractures threaten the health, independence, and survival of millions of people nationwide. An estimated 40% of women and 30% of men will suffer a hip, spine, or wrist fracture in their lifetime. Two million Americans experience a fracture annually, resulting in more than 430,000 hospital admissions, 2.5 million medical office visits, and 180,000 nursing home admissions. Due in part to an aging population, the cost of osteoporotic fracture-related care will exceed $25 billion by 2025. This suffering and cost is preventable. Large randomized controlled trials (RCT) have demonstrated the substantial benefit of osteoporosis drug therapies (ODT) in reducing the risk of osteoporotic fractures. Yet, fewer than 50-80% of patients at risk of fracture will receive ODT and half will discontinue them prematurely. Underuse of, and poor adherence to, ODTs stems in part from the lack of evidence about the effectiveness and safety long-term use of anti-resorptive ODT (i.e., bisphosphonates and denosumab), particularly with respect to rare side effects such as atypical femur fracture (AFF) and osteonecrosis of the jaw (ONJ). While interruption of long-term ODT (‘drug holiday’) and use of sequential therapies (i.e., anabolic ODT followed by anti-resorptive ODT) have been proposed as ways to limit risks of AFF and ONJ, it remains unknown whether these strategies actually reduce risk of harm without compromising ODT effectiveness with respect to fracture prevention. We will address these knowledge gaps by emulating a series of target RCTs examining the comparative effectiveness and safety of ODT regimens with respect to fragility fractures (primary outcome), AFF, ONJ, and other safety endpoints. We will use claims and electronic health record data from OptumLabs Data Warehouse, a dataset of privately-insured and Medicare Advantage beneficiaries, linked to a 100% sample of Medicare fee-for-service claims, to allow for an unprecedented evaluation of ODT over time and across populations, geographies, health systems, and health plans. We will emulate the following target RCTs (eRCTs): Aim 1) eRCT 1 comparing ≤3 years vs. >3 years non- interrupted anti-resorptive therapy. Aim 2) eRCT 2 comparing non-interrupted long-term biphosphonate ODT (>3 years) vs. short-term (≤3 years) ODT followed by either brief (≤3 years) or prolonged (>3 years) drug holiday. Aim 3) eRCT 3 comparing sequential therapy with anabolic ODT followed by >3 years bisphosphonate vs. denosumab treatment. Within each eRCT, we will assess for heterogeneity of treatment effects as a function of gender and risk profile (e.g. steroid use, etc). Finally, to address the gaps in translation of evidence to patient care decisions, Aim 4 will update and field test an effective but outdated shared decision-making tool (Osteoporosis Choice) with data produced by Aims 1-3. At the conclusion of this work, we will generate both the evidence to inform high quality osteoporosis care and a ready-to-use shared decision-making tool to support the implementation of this evidence into practice to improve the health of patients with osteoporosis.

PROJECT NARRATIVE Using real-world observational data, we propose to emulate three randomized trials comparing the benefits and harms of osteoporosis drug regimens in adults at a wide range of risk for fragility fractures. Furthermore, we will produce a ready-to-use shared decision-making tool to translate the knowledge produced into practice. 1