Using Real-World Data to Enhance Clinical Trials

Clinical trials are generally considered the foremost authority for generating robust medical evidence because of their methodological strengths relative to other clinical research designs. However, they are susceptible to substantial challenges, such as enrollment barriers, low participation rates, high operational costs, and limited results generalizability, to name a few. A promising resource to address these challenges is real-world data (RWD), generally defined as routinely collected data during the delivery of healthcare. Database-specific RWD – such as electronic health records (EHRs), administrative claims, and clinical registries – is of particular interest for their richness and volume. However, coordination between the primary data collection actions of clinical trials with the secondary collection nature of RWD, while also accounting for data fitness-for-use considerations, persists as a prominent challenge. This dissertation aims to advance the sciences of using RWD to enhance clinical trials, specifically from two perspectives: (1) a trial design perspective; and (2) a results interpretation perspective. It first reviews relevant literature about RWD uses for clinical trial conduct. It then seeks to address two research questions focused on using RWD to improve clinical trials, with particular emphasis on clinical trials that evaluate medications: (1) how do eligibility criteria, both individually and in combination, affect patient safety and recruitment pool size; and (2) how representative of real-world patients are enrolled trial participants. The utility of RWD in investigating these questions is tested using two aims. Aim 1 examines the impact on hospitalization risk and eligible patient pool size of different eligibility criteria combinations across a variety of disease domains. Aim 2 clinically characterizes trial participants for generalizability assessments. The primary innovations of this dissertation include (1) supplementing a RWD source with trial enrollment data, thus creating a novel combination for enriched evaluations; and (2) developing innovative approaches, both across sets of clinical trials and within individual trials, for generalizability assessments. Ultimately, the findings of this dissertation demonstrate how clinical trial design, and the interpretation of their results, can be enhanced through the use of RWD in order to strengthen clinical research pursuits in study design and results interpretation.