Developing Statistical Tools and Visualization Methods for Understanding Heterogeneity in Distributed Networks: Applications to COVID-19 and Diabetes

Scientists at the Canadian Observational Network for Drug Effect Studies (CNODES) have created ways to combine results from separate Canadian provinces, the United Kingdom, and the United States to rapidly assess the safety and effectiveness of drugs while minimizing the risk of accidentally exposing patient data or jeopardizing patient privacy. Because each province differs from the others, the United States, and the United Kingdom, the best research strategies (as well as the overall safety and effectiveness of drugs) can be very different as well. In 2022, these differences are larger than ever thanks to the ongoing COVID-19 pandemic substantially altering the types of drugs patients start taking and whether they attend in-person office visits with their providers. Currently, however, there are very few ways to understand, document, and communicate these differences. This research project aims to develop new ways for researchers to understand, interpret, and communicate differences in how parts of networks like CNODES (or other studies involving multiple populations) decide which patients receive what treatment, differences in rates of important risk factors and health outcomes, and the extent to differences between populations change research findings and alter treatment effectiveness. These new tools will be tested on data from the United Kingdom's Clinical Practice Research Datalink (CPRD), treating six geographic areas as if they were each contributing data to a distributed network like CNODES. Finally, the tools will be used to study 2019, 2020, 2021, and 2022 data from the CPRD, Ontario, and British Columbia to understand how the timing of lockdowns and other pandemic prevention measures impacted whether patients initiated various drug treatments for high blood pressure, diabetes, and depression, as well as the frequency of various important health outcomes.