Chronic obstructive pulmonary disease (COPD) is a leading cause of morbidity and mortality worldwide. While numerous scientific advances have been made to improve the care of patients with COPD, current therapeutic options remain limited. The growing availability of electronic health data and new computational tools have brought an unprecedented opportunity to accelerate the discovery of new therapeutic options for COPD via the analysis of existing health care data. We hypothesize that observational health data can be utilized to identify drug repurposing opportunities for COPD in a valid and expeditious way. The overarching goal of this proposal is to develop a population-based analytics and data infrastructure that will allow robust and valid evaluation of non-COPD drugs and COPD outcomes in patients with COPD, leveraging existing claims data sources and state-of-the-art methods for causal inference. The claims data will cover U.S. commercially insured individuals and Medicare beneficiaries. Electronic health records (EHR) and longitudinal data from COPDGene, a cohort of almost 5,000 smokers with longitudinal collection of extensive clinical and patient-reported data over 10-year period, will be linked to claims and used to develop, validate, or further refine key definitions for claims-based analyses, including endpoints, COPD severity markers, and COPD subtypes. The infrastructure will be developed using statins and diabetes medications, and further refined via emulation of selected randomized controlled trials (RCTs) that evaluated non-COPD medications for the prevention of COPD exacerbations in patients with COPD over the past decade. We will also evaluate the comparative effectiveness of new immunomodulating agents, such as dupilumab. While many previous observational studies of repurposed medications have been found to have substantial bias, we will build on the lessons learned in pharmacoepidemiology over the past decade and use causal inference methods that have demonstrated high validity in previous investigations and RCT emulation projects. Finally, as part of the project we will conduct thorough assessment of treatment effect heterogeneity across patient subgroups, including, but not limited to subgroups based on age, sex, race/ethnicity, comorbidities, and comedications, as well as COPD subtypes developed as part of this proposal. The proposed work will generate highly needed evidence on the impact of selected non-COPD medications on COPD outcomes. More importantly, it will form the foundation for future monitoring and evaluation of repurposing opportunities in COPD, expediting the path to discovery of new therapeutic targets for this debilitating disease.

Chronic obstructive pulmonary disease (COPD) is a leading cause of morbidity and mortality in the US, yet therapeutic options are limited. The advent of electronic health data has brought an unprecedented opportunity to accelerate drug discovery and identify new therapeutic options for COPD among existing drugs (drug repurposing). Building on state-of-the-art methods for causal inference and existing claims data sources, supplemented with EHR data and deeply phenotyped data from COPDGene, we will develop, test, and implement an analytics and data infrastructure to efficiently evaluate non-COPD medications for COPD.