Highlights from the Spring 2018 Informatics Showcase, Part 2

Well, we're back after our brief (non-)commercial break! Are you ready to explore some more informatics projects with me? Then get ready as we dive back in with our bite-size summaries of the final three project presentations…and if you're a tad confused or joining us for the first time, check out Part 1 to get caught up!

USING THE INFORMATICS TOOL EMERSE TO IDENTIFY CASES OF DEEP VEIN THROMBOSIS

In Identifying deep vein thrombosis cases using EMERSE, Carlton Moore, MD, MS, Professor in the Division of Hospital Medicine, discussed a hospital quality improvement (QI) project using Electronic MEdical Record Search Engine (EMERSE), an informatics tool used to search free-text clinical notes, to identify cases of deep vein thrombosis (DVT) and pulmonary embolism (PE). Collectively DVT and PE are known as venous thromboembolism, or VTE.

DVT and PE are primarily a problem of bed-bound patients, especially post-operative patients, and approximately 500,000 patients in the US are affected every year. It's also among the most common preventable causes of hospital death. The ultimate goal of this QI project was to reduce hospital-acquired cases of VTE at UNC Hospitals. Anticoagulant prophylaxis treatment for at-risk hospital patients can reduce VTE by 30-50% with a low incidence of major bleeding. Unfortunately, only about 60% of surgical inpatients receive appropriate prophylaxis.

Prior to EMERSE, there was no way for hospital quality improvement teams to identify acute VTEs in real-time because the diagnosis is buried in free-text clinical reports. QI teams had to rely on retrospective billing data to perform chart reviews and root-cause analyses. These were often not accurate and they certainly weren't timely since they couldn't be done until after patient discharge.

Using EMERSE it was possible for Moore's team to set-up an automated system that accurately identified cases of acute VTE in UNC Hospitals in near real-time by searching free-text clinical reports, including lower extremity ultrasounds, ventilation perfusion (V/Q) scans and chest CT scans. This allowed the team to discuss identified cases with the patients' providers and pursue more accurate root cause and risk analyses than were possible previously. This led to the development of three QI projects to address issues identified that will hopefully result in fewer cases of hospital-acquired VTE at UNC Hospitals in the future.

My takeaway from this presentation – EMERSE, sometimes called the Google for free-text clinical notes, is pretty darn awesome.

USING ELECTRONIC HEALTH RECORDS TO INVESTIGATE EFFECTS OF AIR POLLUTION EXPOSURE ON HEART FAILURE PATIENTS

In Effects of air pollution exposure on heart failure patients as revealed by electronic health records, Cavin Ward-Caviness, PhD, Computational Biologist with the Environmental Protection Agency (EPA), discussed a project designed to examine the link between long-term air pollution exposure and mortality in people with heart failure. The study was conducted in collaboration with the TraCS Informatics and Data Science (IDSci) Team utilizing data from the Carolina Data Warehouse for Health (CDW-H).

In Meet EPA Computational Biologist Cavin Ward-Caviness, PhD, Ward-Caviness had this to say about the collaboration, "We are right now using it to study air pollution effects in heart failure patients – a very understudied population who have shown high sensitivity to air pollution exposure. We are calling this effort EPA CARES, and I think it has the ability to be a great resource to let us look at health risks due to environmental exposures and community characteristics in the general population and particularly in understudied population groups."

I'm pretty sure that the data presented at the showcase has yet to be published, so I'm not going to go into other details. Definitely keep your eyes out for publications from the project in the future!

My takeaway from this presentation (no spoilers here!) – The EPA has a lot of monitoring stations throughout North Carolina! Also, the intersection of the air quality data and EHR data is quite powerful & really brings home the effects of air quality on health.

EXTRACTING THE ACTIONABLE EJECTION FRACTION FROM FREE-TEXT MEDICAL RECORDS

In From free-text to structured data: Extracting ejection fraction from clinical notes, Larry Klein, MD, Professor of Medicine and Radiology, shared an ongoing project to identify and extract ejection fractions from free-text clinical notes, and then convert that information into structured data that would be both more accessible and easier to pull together for various reports and registries. [Say what??? Keep reading – I promise it'll make more sense by the end!]

Ejection fraction is a measurement of how much blood is being pumped out of the heart each time that it contracts, often referring to left ventricular ejection fraction. Why is this important to know? Knowing a patient's ejection fraction is critical for cardiac patient care – it's used to determine medications, to decide if implantable devices (e.g. implantable cardioverter defibrillators) are needed, and it helps identify high risk patients who are more likely to be readmitted to the hospital or to experience sudden death due to arrhythmias.

Ejection fractions are also determined in a lot of different ways, from a myriad of procedures with different reporting formats, and using different data capture methods. There's also a difference between the calculated ejection fraction and the ejection fraction reported by the physician. The ejection fraction that is considered actionable is the one that the physician reports, and this ejection fraction is most often found in free-text clinical notes.

Klein and his team created a program that parses the data from myriad locations and builds a database with the actionable ejection fraction for the Carolina Data Warehouse for Health (CDW-H) and populates a tab in Epic for heart failure patients that includes all of their ejection fractions.

My takeaway from this presentation – A task that to a non-clinician, non-informatician seems both straightforward and simple is actually neither simple nor straightforward…AND critical for patient care.

And that's a wrap for our round-up! I hope that, like me, you've got a better sense of the breadth of the Informatics and Data Science fields as well as what our IDSci Team does here at the NC TraCS Institute.

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

Next up – wouldn't you like to know! ...so would we...