Imaging Informatics: EHR Search Tools
By David Yeager
Vol. 15 No. 9 P. 6
The capability to “google” record data will prove valuable to radiologists and other physicians.
EHRs have dramatically increased the amount of information available to medical practitioners, but they also present a conundrum: How do you find the information you need in a sea of data? Software that aids searching is becoming a must-have tool. In addition to improving the quality of care, EHR search tools can be deployed in various ways, and they have the potential to help reduce overutilization of imaging exams.
A study published in the June issue of the Journal of the American College of Radiology looked at 74 common parameters that emergency department physicians deemed important for guiding a patient’s care. Using a software-based search tool, the EHR was queried for various aspects of patients’ medical history, including whether they’ve had heart problems, suffer from diabetes, or have a history of ectopic pregnancy. The mean search time was 15 seconds, with a margin of error of five seconds.
“With the onslaught of information that everyone’s dealing with, there’s been a greater push and increasing evidence that shows decision-support tools can minimize unnecessary care and optimize the care the patients receive,” says Arun Krishnaraj, MD, MPH, a radiologist at the University of Virginia in Charlottesville. “These kinds of decision-support intelligence can be built into a software platform that can then expedite patient care but also provide a much greater level of safety because it allows the busy and often distracted physician to make sure they don’t miss something important.”
The software that was used in the study is called Queriable Patient Inference Dossier (QPID). It’s able to search structured text, such as lab values, as well as unstructured text, such as physician notes. Before moving to the University of Virginia, Krishnaraj was part of the team that developed QPID at Massachusetts General Hospital in Boston. He has done a significant amount of research on the tool and its uses, including a yet-to-be-published study that found a 20% reduction in radiologist interpretation time when using QPID compared with working with a traditional EHR platform. He says tools like QPID can also provide decision support to help reduce imaging overutilization.
Like Google for the EHR
QPID allows users to enter multiple search terms, greatly reducing the time spent scrolling through documents to find relevant information. In addition to directed searches, it allows physicians to create various dashboards that provide commonly needed information automatically, before they start dictation. For example, a radiologist looking at a prostate MRI scan can set up a dashboard that indicates whether the prostate has been biopsied, the patient’s Gleason score, the results of the patient’s digital rectal exam, notes from the urologist, all prior prostate imaging, and a list of other risk factors, such as prior surgeries or radiation exposure, which may affect what’s seen on the images. If any of those biomarkers are present, they will appear in bold; if not, the dashboard grays them out. Parameters that are significantly abnormal appear in red. The radiologist can also view the results of all previous searches run on the patient.
The University of Virginia is currently using QPID on a trial basis in its radiology, cardiology, and internal medicine departments. The organization launched QPID with a radiology-specific dashboard split into specific divisions and different modalities. Although its use is in the early stages at the University of Virginia, Krishnaraj says, the pattern resembles what happened at Massachusetts General Hospital: QPID use was buoyed by trainees with volumes of studies to read, then filtered up to senior staff.
“Once people get up to speed on the interface and the functionality, the more tech-type people start to write their own queries—the barrier to entry for that is very low—and start coming up with really innovative, neat ways to use it. Even novice users, once they get used to it, start to feel that it’s an integral part of their workflow,” Krishnaraj says. “It effectively becomes like a Google for the EHR so, when you’re searching for any piece of information, even if you just want to do a direct word search, just type it in and it’s there.”
Krishnaraj says the software is not difficult to install and can be up and running in a few weeks. The biggest issues are the IT resources that a health care organization has available and getting approvals from administration. The University of Virginia uses Epic’s EHR and whether it will continue using QPID depends, in part, on how much search functionality Epic includes with the next version of its EHR, which should be released this year.
Krishnaraj believes applications for EHR search tools will expand beyond answering immediate clinical questions. He says such tools can also be used to do population health studies or help answer broader research questions. For example, he’s working with endocrine surgeons to determine how many patients, from a group that had incidental adrenal lesions on imaging, had no follow-up appointments within 18 months. It may also be possible to see how large groups of patients respond to certain therapies or medications. The American Recovery and Reinvestment Act and the Affordable Care Act took steps to incentivize health information exchange, he says, but to further these types of efforts, vendors and health systems need to make health information easily accessible and exchangeable, with proper security in place.
“Once you get to that, tools like QPID can be very powerful in looking at exactly those types of issues in population health,” Krishnaraj says. “You now have a tool to ask questions about population health, as long as health information systems are shared, in some manner. [There are ways to] blind the information that is being shared so you can’t identify individuals, but you can still search and do Big Data analysis on population health matters.”
— David Yeager is a freelance writer and editor based in Royersford, Pennsylvania. He writes primarily about informatics topics for Radiology Today.