Skip to content Skip to navigation

How Unleashing Trapped Clinical Data Has Saved Lives at Geisinger Health System

February 17, 2016
by Rajiv Leventhal
| Reprints

Note: Geisinger’s project around treating abdominal aortic aneurysm was named a semifinalist in the 2016 Healthcare Informatics Innovator Awards Program. Short descriptions of the projects of all of the semifinalists in this year’s program can be seen here.

Not too long ago, Gregory Moore, M.D., Ph.D., neuroradiologist and chief emerging technology and informatics officer at the Danville, Pa.-based Geisinger Health System, suspected that abdominal aortic aneurysm (AAA) patients may not have been receiving their necessary follow-up care. While radiologists identify AAA in diagnostic reports, the potential exists for the report to file in the ordering physician’s electronic health record (EHR) with the message regarding needed follow-up missed, or at least not digitally actionable.

As such, Geisinger convened a multi-disciplinary team to implement an end-to-end process that provides a safety net and improves follow-up care for patients with this life-threatening condition—one which James Elmore, M.D., director, endovascular services at Geisinger, calls the “silent killer.” Dr. Elmore says, “We call it that because it’s in the abdominal area and many people don’t even know they have it. Often, they have to get screened for it or it’s found incidentally when getting tested for something else, such as getting a CT scan or ultrasound for another problem.” Nonetheless, Elmore says, AAA is indeed scary, ranking as the 14th leading cause of death in the U.S., affecting 10 percent of men age 65 or older. “You would probably be surprised by that stat and that’s because other medical conditions get more publicity,” Elmore says.

According to Elmore, a patient’s AAA condition will get put in his or her medical record or radiology report, but when it’s put in, the aneurysm might be small in size meaning no immediate treatment is needed. But then as years go by, the aneurysm grows larger, and as it gets bigger, it becomes a serious medical condition, he explains. “So when family doctors get these reports, it might be a small aneurysm to begin with, meaning you don’t have to worry about it this especially when the patient might be facing other serious conditions. But many years go by and all of a sudden, now it’s a big problem. We wanted to make sure that these patients’ aneurysms didn’t get lost in the shuffle when another problem was being dealt with,” Elmore says.

James Elmore, M.D.

As such, the Geisinger team reviewed industry best practices, including the Society of Vascular Surgery practice guidelines for the care of patients with AAA. The guideline categorizes aneurysm size into three groups to appropriately manage outreach. It is reported that larger AAAs tend to grow more rapidly than smaller aneurysms and have a higher rate of rupture, therefore active monitoring is needed as the aneurysm progresses, Geisinger officials noted.

To address the challenge of identifying essential AAA information in unstructured radiology reports, Geisinger introduced natural language processing (NLP) into this effort allowing its team to identify those cases where the information only exists in the narrative radiology report. The finalized Geisinger guidelines were used to develop the clinical decision support necessary to identify, classify and facilitate reliable follow-up care. As a result, the Close the Loop Program allows Geisinger to proactively outreach to patients and physicians where it finds that no follow-up is evident. Elmore adds, “We leverage the data, including unstructured data, to drive interventions and provide clinical decision support.”

Joan Topper, vice president of clinical informatics at Geisinger, organized the team that designed and implemented the AAA Close the Loop Program. The team used the guidelines and NLP to analyze more than two million radiology reports and identify patients with a confirmed AAA, monitor size progression, and ensure patients are not lost to follow-up. What’s more, physician office schedules are accessed to monitor follow up by primary care physicians (PCPs) and specialists. Nurses responsible for proactive outreach are accountable to contact the patient, the PCP or the ordering provider to recommend follow-up when a care gap is identified, Topper notes. “The healthcare industry has used NLP in isolated instances, but very few organizations appear to have integrated its use into clinical operations” says Topper. “Using this technology to data mine the unstructured text document, and using it to improve the data in the EHR to put it on the patient’s problem list as a discrete finding that is far more prevalent to the care team when they see the patient a year or two later has really helped with the reliability in ensuring that follow-up care,” she says.

Geisinger also resolved a data quality issue affecting follow-up care reliability, notes Topper. EHR analysis showed that 61 percent of confirmed AAA patients had the condition listed on their active problem list. Protocols were implemented to provide ongoing updates to problem lists (where feasible) for all unrepaired AAA cases. This improved EHR data quality and patient monitoring by increasing visibility of the condition to the care team. Geisinger studies found that physicians are 90 percent more likely to monitor a condition if identified on the active problem list, Topper says.