March 2015

Right Tool, for Which Job? Radiologists Seek Better Ways Utilize EHRs in Imaging Workflow
By David Yeager
Radiology Today
Vol. 16 No. 3 P. 10

Ever since PACS first came online, integration with EMRs has been a goal of the radiology community, says Eliot Siegel, MD, a professor and vice chair of information systems at the University of Maryland and chief of imaging for the VA Maryland Healthcare System in Baltimore. The Baltimore VA went live with the world's first PACS system in 1993. Since then, the goal of more efficient integration hasn't changed, but what has changed is radiology's concept of what that entails. As EMRs have evolved into EHRs, both the types of information they store and the potential value of that information for clinical practice have grown exponentially.

"As time as gone on, our expectations for the degree of interface between the EMR and the radiology and PACS systems have only become greater and greater," Siegel says. "I'm absolutely convinced that we can become a safer department, a more efficient department, and provide better patient care, as we develop greater integration with the EHR."

Early on, radiologists wanted the reason for the exam, lab values, allergies, and a modality worklist. They now have access to those items, but there is much more information available that could add value to radiology reports. Knowing the patient's smoking history, family history, and history of other illnesses adds context to the patient's condition and helps radiologists make more sensitive and specific interpretations. Many health care providers are interested in tapping into that information, but much of it is not yet incorporated into radiology's workflow.

"For us, there's a lot of pressure to try to pull more and more content from the EMR into PACS and worklist reporting," says R.L. "Skip" Kennedy, MSc, CIIP, technical director of imaging informatics for Kaiser Permanente medical centers in northern California. "So that's really our push, and the mandate we've been given is that there's a number of critical clinical context information points that we need to pull into those systems, but [the EHR] hasn't, so far, been that much of a game changer for the radiologists. And we have rads who never use the EMR."

(Lack of) Ease of Use
Kennedy says EHR users tend to break along generational lines. Younger users who used an EHR/EMR system in medical school are more likely to use it now. Older radiologists tend to view it as an inconvenience. Kennedy thinks natural attrition will continually tip the scales toward EHR users but, even for those who are comfortable in the EHR environment, time spent searching for information reduces productivity.

"Even if you're agile with [the EHR], going out of your PACS context to go get a creatinine is inefficient, in and of itself," Kennedy says. "You may be good at it, you may be fairly fast, you may say it only takes you 45 seconds to navigate to the right place and then get back. I would say that better than having it in 45 seconds is having it literally looking at you in the PACS context. To me, that's the ideal efficiency."

Kaiser Permanente's EMR vendor makes a significant amount of data available through Web services, but it still requires both time and manpower to set up the interfaces that will route the data. Even then, making that information available directly to radiologists will alleviate some of the workflow issues, but it won't completely address them; the volume of information is too great. Kaiser Permanente's radiologists compiled a list of about 40 items that they would like to see. Kennedy says the list will most likely be whittled to around 10. He says the goal is to make the most sought-after items available in the radiology reporting system so they are automatically present when the radiologists issue their reports.

For practices that contract with multiple hospitals that use disparate EHR systems, navigating patient records can be even more daunting. Practices that do a significant amount of screening have to read images and render an interpretation in a very short time window, all day. Robert L. Klein, practice administrator for University Diagnostic Medical Imaging in New York City, says, although EHR information is valuable for radiologists who perform consultative services, there isn't as much value for screening radiologists.

"Is the world expecting a radiologist who has access to five different EMR systems to search for that patient five times to learn about the patient's history, when [he or she is] doing a simple CT of the abdomen for pain?" Klein asks. "And, if that patient comes back two years later with cancer and there was a small, evidentiary piece of information in one of those five different EMRs that might have swayed the radiologist's recommendation, is he really culpable to do all that work?"

Klein notes that some radiology groups who read for hospitals only get paid the reading fee, not the significantly higher technology fee. Radiologists in these types of practices need to read around 50 cases a day, which leaves them about 7 minutes per patient. In such an environment, extra seconds spent searching the EHR can add up fast. Klein says many radiology groups only look at the information they gather directly from the patients and don't search for more. Until EHRs are able to better share information, such as family history and information related to the effects of disease treatment, Klein thinks their value for radiology will be limited.

"When we finally get fully electronic transmission and sharing, and we can see some other vital information about a patient's history, EHRs will be helpful, from the radiologists' viewpoint, with correlating diagnoses on difficult cases because we'll have better background information," Klein says.

Too Much of a Good Thing?
That's not to say that radiologists haven't benefited from EHR information. At the University of Texas M.D. Anderson Cancer Center in Houston, radiologists have one-click access to pathology reports, lab reports, and other clinical documents stored in the EHR. Kevin W. McEnery, MD, a professor of radiology and director of innovation in imaging informatics at M.D. Anderson, says he doesn't allow his residents or fellows to use the phrase "clinical correlation is suggested" in their reports because they have access to the patient's clinical information. If the information they need isn't in the clinical record, they can query the referring physician. He says this practice is one way that radiologists can demonstrate value to referring physicians.

In addition, M.D. Anderson radiologists are able to read exams based on when a patient's clinical appointment is scheduled, rather than simply reading exams in the order they are performed. McEnery says this allows the radiology department to orchestrate workflow more efficiently in their outpatient practice, by making a higher percentage of studies available for the appointments. He adds that the set-up of the EHR interface has a significant effect on workflow efficiency and influences utilization rates. He also sees a need to take PACS/EHR integration a step further.

"I think what is necessary are ways to extract the data that fit into the radiologists' workflow. In general, if you look at the EMRs, the clinicians are documenting information on the patient's clinical care in a way that should allow the radiologist to extract the clinical history. The issue is how efficiently they can extract that clinical history and move on with their interpretation," McEnery says. "What you will probably see is more and more focus on information displays that not only allow the radiologists to be as efficient as they currently are, but also have more ready access to clinical screens that automatically display the information, without the radiologist having to go into the patient's record and search for data in different locations."

Siegel agrees that, even with a relatively user-friendly interface, managing an open EHR screen while reading studies in PACS can be problematic. Because of security and privacy features in many EHRs, they will frequently sign off automatically, requiring the radiologist to sign back in for every different patient. Also, most EHR systems function as digital document systems: Stored information can be retrieved but not queried. Siegel says the ability to query the EHR would save a lot of time.

"If I want to call up a patient's progress note or discharge summary from a particular date, I can call it up and read it, almost as though the EMR were just a digital document system," Siegel says. "But, if I want to do a search, for example, if I want to search for all of the instances in which my patient presented with a circular maculopapular rash, most of the EMRs have not been designed to allow me to search through a particular patient's records, much less search across different records."

Although people are developing mechanisms that can query the EHR, such as a Health Level 7 International (HL7) aggregator that monitors HL7 messages and submits requests for reports to the EHR, there is currently no mechanism that allows a direct EHR database search. Siegel believes more efficient EHR querying, similar to the way people use Web browsers, is an important step toward better EHR/PACS workflow. Beyond that, he would like to see an intelligent interface that can learn what information he needs, depending on the types of studies he's reading, and deliver it automatically. This type of function is often called a dashboard.

Siegel also sees room for growth in the functions that EHRs may potentially enable, such as better tracking of patient radiation information and abnormal findings; communication of preliminary findings from the emergency department; information about the quality of radiology studies; and better correlation of radiology findings with other data, such as laboratory, clinical, pathology, and genomic findings. He's currently working on a project that can identify a lung nodule at a PACS workstation; examine the characteristics of the nodule; check the relevant patient history, such as smoking history, age, sex, geographic location, and ethnic background, in the EHR; correlate information about the nodule with the patient history; and make a prediction about the likelihood of the nodule being malignant based on database information, such as the data in the National Lung Screening Trial. That sort of interplay between EHRs and imaging systems would greatly advance decision support capabilities.

Siegel says there will be an increasing number of decision support algorithms available in EHRs. He adds that a continuum of communication between EHRs and PACS would not only help radiologists produce better interpretations, but would also help referring physicians make more effective treatment decisions and, potentially, be a step toward personalized medicine. For example, an incidental finding of coronary artery calcification or a compression fracture of a vertebral body may influence a physician's decision to prescribe calcium or therapy for low bone mineral density, he says.

Decision support is another area where radiology can demonstrate value to referring physicians and improve patient outcomes. In this era of changing reimbursement structures and value-based care, those are essential considerations. Despite the current inconveniences, McEnery says radiologists will have no choice but to actively incorporate EHR data into their interpretations.

"There are going to be groups who are going to aggressively leverage the EMR to aggressively learn about the patient's history," McEnery says. "And, over time, I can assure you that, if clinicians have a choice between imaging centers, they're going to migrate to the imaging practices that they believe provide them with the most clinical context and, therefore, the chance for a better clinical outcome for their patients."

— David Yeager is a freelance writer and editor based in Royersford, Pennsylvania. He writes primarily about imaging informatics for Radiology Today.