Despite almost complete absence of adequate data from head-to-head randomized controlled trials, treatment guidelines and prescription drug insurance formularies typically consider individual drugs within medication classes as equally effective and equally safe. Yet, incorrect assumptions regarding therapeutic exchangeability expose patients to suboptimal treatments and adverse clinical outcomes, particularly older adults, who are disproportionately affected by chronic conditions and are the largest per capita consumers of prescription medications. Despite its substantial clinical and economic implications, therapeutic exchangeability remains remarkably understudied and represents a problem without a feasible current solution. We thus propose a novel and feasible approach to evaluate the therapeutic exchangeability of same-class drugs. The proposed studies will take advantage of natural experiments created by the structure of the Medicare Part D drug benefit and the variable financial incentives that Part D plans receive from manufacturers. Due to plan- specific formulary management strategies, Part D enrollees initiating a new medication often face substantially different out-of-pocket costs for alternative drugs within the same class. The differences in out-of-pocket costs among alternative same-class drugs among the hundreds of Part D plans will serve as instrumental variables (IVs). Because these financial incentives strongly affect the choice of one drug of a class over another and are independent of the patients’ clinical characteristics (as demonstrated by strong preliminary data), they facilitate valid IV estimation. Outcome validation from primary medical records and cause of death data from the National Death Index further improve the rigor of the study. Using existing data on >22 million Medicare beneficiaries, the proposed study will examine 4 carefully selected drug classes to establish a new methodological framework for the systematic assessment of within-class therapeutic exchangeability from observational data: 1) direct oral anticoagulants (DOACs) for stroke prevention in atrial fibrillation or atrial flutter, 2) dipeptidyl peptidase 4-inhibitors (DPP-4s) for type 2 diabetes, 3) high potency statins for secondary prevention of atherosclerotic cardiovascular (CV) disease, and 4) sodium-glucose co-transporter-2 inhibitors (SGLT-2s) for type 2 diabetes. These were selected based on explicit criteria: high rates of use in Part D beneficiaries, uncertainty about therapeutic exchangeability within the class, sufficient variation in out-of-pocket costs among alternative agents, and ability to validly measure outcomes in Medicare claims. We included examples with strong priors against (DPP-4s and CV outcomes) and for (SGLT-2s and amputations) within-class differences to show we can reproduce expected findings, and others for which differences are uncertain (e.g., DOACs and ischemic stroke). This proposal begins a highly promising novel line of work to feasibly generate valid and critically needed evidence on therapeutic exchangeability within widely-used drug classes among older adults, serve as the basis for future confirmatory studies, and improve clinical medicine, patient outcomes, and public health.

Despite the absence of supporting evidence, treatment guidelines and prescription drug insurance formularies typically assume that individual drugs within medication classes are equally effective and equally safe. Yet, when incorrect, these assumptions expose patients to suboptimal treatments and clinical harms, particularly older adults, who are disproportionately affected by chronic conditions and take more prescription medications than any other population. Taking advantage of natural experiments created by the design of the Medicare prescription drug benefit, the proposed study will use existing data on more than 22 million Medicare beneficiaries to answer key questions about the comparative benefits and harms among drugs within each of four therapeutic classes widely-used among older adults.