Adapting epidemiologic principles to “big data”

Optum’s epidemiology experts apply the rigor of advanced epidemiologic methods to the challenges inherent in “big data”

David DoreThe term “big data” is ubiquitous these days. Within healthcare, electronic data from administrative systems and electronic health records (EHRs) are increasingly voluminous, varied and accessible. Within health care and public health, our capacity to use data has not scaled as quickly as the data themselves. Data management, processing and rapid analytics solutions are emerging and necessary to fulfill the promise of “big data.” Rapid analytics will improve the delivery of health care, especially if findings are available to stakeholders in a timely fashion.

Optum’s epidemiology experts embrace “big data” and emerging analytic tools. However, these innovations are insufficient. At Optum, we are rapidly adapting epidemiologic methods to “big data” from health plans (claims) and providers (EHRs) to characterize populations and the outcomes resulting from exposure to medical products.

We shared our recent developments via contributions to the 31st International Conference on Pharmacoepidemiology & Therapeutic Risk Management (ICPE) last week in Boston. Optum’s epidemiology team specializes in pharmacoepidemiology, which is the application of epidemiology to the study of medical products and involves observing and comparing carefully-chosen populations of patients rather than individual patients. The conference is sponsored by the International Society for Pharmacoepidemiology (ISPE), the premiere society for pharmacoepidemiology in the world, and our epidemiology team participates each year to learn alongside fellow thought leaders, researchers and industry peers.

The explosion of rich health care data presents substantial promise for epidemiologic research. However, these big data are complex, heterogeneous and come from various sources – presenting many challenges. Solutions to these challenges require understanding, managing and determining appropriate applications of the data, all of which are helped when viewed through the lens of epidemiologic principles. There is little doubt that the health system will learn from rapid summarization and visualization of data. Nevertheless, more complex tasks such as those involving characterizations of cause and effect require methodological and subject matter expertise that exceed the capability of general “big data” solutions.

At Optum, we are actively working on innovative data linkages, study designs and analytic methods to conduct sophisticated research with “big data.” The fields of health care and public health must go beyond analytics – indeed must go back to the basics of good epidemiologic practice and then forward to the realization of “big data’s” potential. At ICPE we presented a symposium on specific examples of adapting a large EHR for formal research and appreciated the chance to engage with the wider community on our progress.

Thank you to ICPE for an outstanding conference and to our industry counterparts who attended and participated. These types of forums facilitate learning and discussion, fostering new forms of collaboration that are vital to advancing the initiatives we’re working on and helping us to reflect on current goals while refining our path forward. To learn more about our group and our work, visit us at

–David Dore

Leave a Reply