The COVID-19 Research Database: Building One of the Largest PRO Bono Real-World DATA Repositories
Author(s)
Talwai A1, Wing V2, Itzkovich Y2, Galaznik A2, Chatterjee A3, Jain R4, LaBonte J5
1Acorn AI, a Medidata company, Revere, MA, USA, 2Acorn AI, a Medidata company, Boston, MA, USA, 3Acorn AI at Medidata, a Dassault Systèmes company, Boston, MA, USA, 4Acorn AI, a Medidata company, Arlington, MA, USA, 5Datavant, San Francisco, CA, USA
Presentation Documents
Objectives: The creation of an integrated data repository to capture the comprehensive patient journey across multiple data sources has always held promise in principle but has been stymied in practice due to the challenges of managing data provider partnerships, patient privacy, and technological integration. However, the urgent need to better understand the effects of COVID-19 propelled the creation and operationalization of an integrated data repository. The aim of this research is to chronicle the technical and collaborative efforts underpinning this success. Methods: The COVID-19 Consortium is a pro-bono, cross-industry collaboration with academia, comprising of institutions donating technology, healthcare expertise, and de-identified data. Different types of patient level data from several industry-leading sources were ingested through an automated pipeline comprising source-native ETL (auto-refreshed weekly), privacy-preserving tokenization linking patients across datasets, and granular project-based access. The data and analytical environments were hosted in a dedicated and isolated virtual network to protect data privacy and intellectual property. A third-party certifier safeguarded protection of patient privacy while scientific and pan-stakeholder governance ensured research quality, data security, and operational collaboration. Results: In 2020, approximately 9 terabytes of data, consisting of 5 billion records from 250 million unique persons and 2.1 million patients with COVID-19, from ten different data providers were loaded onto the platform. Data types ingested include medical, pharmacy, and life insurance claims, electronic health records, mortality, consumer, and health propensities. Over 350 academic, scientific, and medical researchers accessed the database and produced 23 publications in medical journals and the media, with over 100 ongoing projects. Topics of research ranged from the public health effects to the socioeconomic effects of COVID-19. Conclusions: The successful deployment of the COVID-19 database is a concrete example of the possibility and potential of integrating disparate datasets for rapid research and can serve as a roadmap for future efforts.
Conference/Value in Health Info
2021-05, ISPOR 2021, Montreal, Canada
Value in Health, Volume 24, Issue 5, S1 (May 2021)
Code
PIN83
Topic
Epidemiology & Public Health, Health Policy & Regulatory, Real World Data & Information Systems
Topic Subcategory
Distributed Data & Research Networks, Health Disparities & Equity, Public Health
Disease
Infectious Disease (non-vaccine)