Riding the Technology Wave

Decision support for radiologists rises to the point of care.


For the first hundred years of radiology, the medical chart always accompanied the patient, and the radiologists never saw it. As a result, radiologists reading images often had no knowledge of the patient’s medical history.

Does the patient have diabetes or cancer or some other condition that might affect the findings? In more ways than one, radiologists were practicing in the dark.

With the advent of electronic health records (EHRs), it became possible for radiologists to gain access to vital patient history. Unfortunately, however, the EHR was primarily designed to meet the needs of other physicians, and radiologists were mostly an afterthought.

“Finally, we’re getting the picture that medical records contain a lot of information that radiologists would like to know — if only they could get at it quickly and easily,” says David S. Hirschorn, MD. Hirschorn is chief of informatics at Imaging Service Line at Northwell Health, New York state’s largest health care provider with 21 hospitals and over 550 outpatient practices.

Until recently, for each patient, radiologists had to navigate the EHR and drill down to relevant information. If that discovery process took too long, however, many radiologists — especially when under pressure to speed turnaround time — gave up and simply read the exam.

“The battle (or art) in 2017 is to create smart algorithms that bring radiologists exactly what they need for the task at hand,” says Hirschorn. “For example, if I’m reading a CT of a patient with pancreatitis, please just tell me the amylase and lipase lab values right away. Don’t make me dig for them. It’s all about having the right information in the right place at the right time. That’s what clinical decision support (CDS) for radiologists is all about.”

Accessing Guidelines and Mitigating Errors

Many think of CDS for imaging as strictly applying to decision support for order entry by referring providers. However, decision support tools can aid diagnostic radiologists as well. Beyond putting meaningful information about a patient’s medical history at radiologists’ fingertips, CDS for radiologists also ensures the use of current guidelines and best practices for interpreting images to help make clinical decisions.

“The purpose of the radiologist is not just to say what’s there; it’s to help the referring doctor know what to do about it,” says Hirschorn. “Say you find a pulmonary nodule on a chest scan. Very often it’s benign, but sometimes it’s cancer. An algorithm developed by a team of experts will tell you the best thing to do next depending on the patient’s age, the size and location of the nodule, and other factors like smoking history. In the past, radiologists would try to dig up that algorithm from a book or a sticky note or an online search. Now, they can use CDS to instantly provide a report with the appropriate recommendation.”

Having clinical decision support at the point of care means not only enhanced efficiency for radiologists, but also fewer mistakes. “A big stumbling block in a radiologist’s work is the inability to quickly access the information they need to know along with the pressure placed on them for report turnaround time, which is a metric used by virtually every institution,” says Arun Krishnaraj, MD, MPH, chief of the division of body imaging at the University of Virginia Health System. “When the focus is on getting the report out quickly, sometimes errors creep in or the report is incomplete because the information is just too hard to get.”

Krishnaraj adds, “Decision support puts tools at our disposal that allow us to get the information we need and to produce a high-quality report quickly. Ultimately, the report is the final product of our work. For radiologists to demonstrate our value to stakeholders we must produce highquality reports; otherwise, we’re falling short.”

Delivering Benefits for Radiologists, Referrers, and Patients

Cree M. Gaskin, MD, professor of radiology and medical imaging and vice chair of informatics at the University of Virginia Health System, has begun integrating radiology-centric dashboards into the EHR to help radiologists find relevant information and produce more accurate reports.

“Diagnostic radiologists commonly interpret hundreds of studies per day while creating reports that integrate the imaging findings with patient clinical context and current medical knowledge,” says Gaskin. “We may interpret a study quickly — often in seconds — but it might take several minutes to read information in a chart, and longer if it’s a complex chart. By delivering tailored, relevant patient information, CDS improves our quality in making recommendations, because we’re focused on what’s going on clinically.”

According to Gaskin, the benefit of having meaningful clinical information at your fingertips is twofold: a more useful report for clinicians based on patient clinical context and a potentially better outcome for the patient. “For example, you might be looking at the foot, but if you knew that the issue was at the great toe, you might give that toe an extra look,” he says. “By knowing exactly what’s going on with the patient, you’re less likely to miss a subtle finding.”

Quantitative techniques and image-interpretation algorithms based on machine learning offer additional types of CDS that have the potential to help radiologists more consistently detect findings that might not otherwise be made. Examples include automated coronary calcium scoring, fatty liver detection, bone density determination, and lung nodule detection — all of which may show up on a CT of the chest, abdomen, or pelvis that was done for other reasons.

Driving Adoption of CDS for Radiologists

Like every new technology, CDS for radiologists will take some getting used to before users begin seeing its value. But radiologists should persevere, Gaskin emphasizes. “Radiologists should insist upon having access to integrated decision support technology now,” he says. “But for it to be widely adopted, CDS technology must fit seamlessly into the radiologist’s workflow, or it will be ignored.”

Most important, says Krishnaraj, is for radiologists to ensure they have easy access to relevant patient data. To make this happen, ACR Assistradiologists need to specify to practice leaders, health system administrators, and technology vendors exactly what information they need. “At a minimum, we should design structured reporting templates,” says Krishnaraj. “Sit down with your referring providers to ensure the reports include the key elements they feel are necessary, and build those prompts into your template. Finally, bring in your EHR vendor to find out how to populate that information automatically in your report.”

Hirschorn says populating report templates with relevant information using standardized language is exactly where the ACR Assist™ decision support tool comes into play for radiologists. (See sidebar for more information.)

“ACR Assist currently has valuable content for pulmonary and adrenal nodules,” says Hirschorn. “And more is being developed every day. Radiologists who have created standardized protocols or decision support algorithms should contact the ACR to get their content populated in ACR Assist. Contributing to that community effort is important. As radiologists, we need to ride this technology wave, not get knocked over by it.”

By Linda Sowers, freelance writer for the ACR Bulletin

Share this content

Submit to FacebookSubmit to Google PlusSubmit to TwitterSubmit to LinkedIn