I started in my role with the Communications Team here at NC TraCS with little to no understanding of 'Informatics' other than a vague association of the term with both information and technology. It would be stretching the truth to say that I understand 'Informatics' today, but I do have a better idea of what it is and what members of our NC TraCS Informatics and Data Science (IDSci) Team do than I did when I started in this role. At least that's my story…and I'm sticking to it!

A key reason for my increased understanding are the Informatics Showcases that our IDSci Team hold several times a year. These events highlight projects that benefit from the expertise of our IDSci Team or use informatics tools developed or deployed by our IDSci Team – providing the concrete examples that help me figure out a new subject or an unfamiliar topic.

Perhaps, like me, you've discovered that when learning about an unfamiliar subject it helps to delve into examples of that subject before exploring more abstract or conceptual materials. Or, maybe you're an informatics person, but you want to know more about our IDSci Team and how their work helps researchers conduct better research. Either way, I hope that you get something valuable from these bite-size summaries of the Spring 2018 Informatics Showcase presentations!

Without further ado, here are the first two projects summaries:

IMPLICATIONS OF GEOGRAPHIC PROXIMITY IN CLINICAL DATA RESEARCH NETWORKS

In Privacy-preserving methods to link patients across datasets and health systems, Ashok Krishnamurthy, PhD, Director of the NC TraCS IDSci Service, discussed de-duplication, or the detection and removal of multiple electronic versions of the same individual's medical records in the Carolinas Collaborative.

The Carolinas Collaborative is a regional Clinical Data Research Network that is comprised of four healthcare system partners (Health Sciences South Carolina, Duke University, Wake Forest Baptist Health, and the University of North Carolina at Chapel Hill) and that collectively has data on their 12+ million patients. All sites within the Carolinas Collaborative are within 220 miles of each other, and there are two sites only 10 miles apart. As you might imagine, this geographic proximity lends itself to patients with records in more than one Carolinas Collaborative health system. As a result, there's quite the need for a process of de-duplicating the data available, as well as a need to protect the privacy of the individuals whose medical records make up that data!

Instead of creating a common Medical Record Identifier that would be used across all four institutions, the Carolinas Collaborative developed a project-based de-duplication process that protects patient privacy. This was accomplished by using a distributed model – each institution works with their own data – and the only data shared across institutions is "un-identifiable" data using a hash algorithm to encrypt patient identifiers for linkage.

My takeaway from this presentation – A lot of care and thought goes into protecting privacy while also working to improve the data available for and included in research through the Carolinas Collaborative and other CDRNs.

RESEARCH RECRUITMENT IN EPIC

In Study recruitment success using Epic's Best Practice Advisories, Loretta Fearrington, a Research Informatics Specialist with the TraCS IDSci Team, highlighted how research teams can find recruitment success with Epic's Best Practice Advisories. Research Informatics Specialists and researchers spend time identifying the patients and the data needed to conduct a specific research study. This is often described as "finding the needle in the haystack." However, with some studies just finding the needle in the haystack isn't enough – there's also a need to find that needle at a particular time point – and that can be even more challenging.

One way to capture the elusive element of time is by using the Best Practice Advisory (BPA) functionality within Epic. How does this work? Patient records within Epic's database are scanned and notifications are displayed or sent out when certain specified criteria are met. This is a great way to find the needle in the haystack when it's happening rather than a day or more after it has happened.

In a study highlighted by Fearrington in her presentation, it was crucial for the study team to identify patients with prior diagnoses of dementia when first admitted to the hospital because that was a window of time when family members were more likely to be available to discuss participating in the study. Using a Best Practice Advisory as part of their recruitment plan was critical to the success of the study. Laura Hanson, MD, MPH, Professor of Geriatric Medicine and the Principal Investigator of the study, provided feedback to the IDSci Team that "we could not have done this study without the screening process and related BPAs."

My takeaway from this presentation – Epic BPAs are an efficient way for research teams to be notified of time-sensitive recruitment opportunities. BPAs are particularly useful for research that would be difficult to conduct otherwise due to the time-sensitive exclusion/inclusion criteria. Oh, and these are a type of Best Practice Advisory that isn't annoying to clinicians…because the advisories go to the study team!

Whew – that was a lot, wasn't it? Learning new stuff is all well and good – and it's one of my favorite things – but my brain hurts a wee bit. So, we're going to pause here for a commercial break (not really – no commercials here!), then finish up with the final three projects in our next post.

Join us for Part 2 as we delve into even more examples from the world of informatics!

See also: NC TraCS Institute – Informatics and Data Science; Epic Research Informatics; It's All About the Data | Research Files Vol 1; UNC Informatics Academic Programs

Coming soon — Join us for Part 2 of our highlights from the Spring 2018 Informatics Showcase projects!