Q&A: How workflows are evolving—and how radiologists can keep up

It’s no secret that workflows in radiology are changing. A number of factors are impacting interpretation volume, while cutting-edge technologies are transforming how radiologists work.

Tessa Cook, MD, PhD, assistant professor of radiology at the Hospital of the University of Pennsylvania and Director of the Center for Translational Imaging Informatics, has been focused on improving workflows and tapping into analytics to increase efficiencies. She’s written and spoken on the topic on numerous occasions, and is currently working with colleague Hanna Zafar, MD, MHS, on the Automated Radiology Recommendation Tracking Engine (ARRTE), a tool to help specify the modality and timing of follow-up for indeterminate and suspicious lesions.

Health Imaging recently spoke with Cook to get her insights on how workflow have changed in recent years and what technologies are essential on the road ahead:

How have workflows in radiology changed in recent years?

Tessa Cook: In recent years, the volumes have increased and those have definitely brought workflow challenges to radiology. One of the things that happened in the last 10 to 15 years has been the introduction of voice-recognition, PACS, virtual elimination of transcription…the majority of folks are trending towards electronic solutions. As part of that, we have been able to accommodate more volume, and the combination of more volume—or electronic solutions enabling more volume leading to the need for more electronic solutions—has really been one of the biggest changes in workflow.

That increase in volume no doubt presents challenges. What are some other hurdles related to imaging workflows?

TC: That extra workload has demanded us to be more efficient and more productive. Now, because there’s increased input at the beginning of the pipeline, if you will, we still need to keep up.

One of the challenges that the more senior radiologists will quote you is they never used to be the report editors. They used to dictate onto tape, since a transcriptionist would make sure [the report] was all correct and send it back to them to sign off. Now, when we dictate “male,” voice recognition may insert “female” into the report. That reliance [on transcriptionists] making sure the report is accurate for sometimes very simple things is now falling on the radiologist. That causes inefficiencies.

Another thing that we spend inordinate amounts of time on is trying to get hold of referring colleagues. Sometimes it is easy and sometimes it is extremely difficult, and it can really throw your workflow and efficiency for a loop if you’re spending time on the phone.

Other challenges include navigating a variety of electronic systems. There may be multiple places you have to look for information so that electronic transition that has given us a lot more information is also adding a number of potentially time-consuming steps to the job that we do.

On the topic of technologies, some may be increasing time burdens, but others are essential to efficient workflows. What capabilities are absolutely needed today?

TC: Customizable work lists. It’s pretty easy to say that if a study is ordered stat, it should be put at the top of the list. If it is not ordered stat, to put it below the stat cases. But there’s much more nuance and complexity to that.

For example, we at Penn are a level I Trauma Center, so we have our trauma cases, we have our emergency department cases, we have our inpatient cases—which could be stat cases or not—and then we have our outpatient cases. Even with our outpatient cases, you have patients that have doctor’s appointments later that day coming in the morning for imaging; their study needs to be interpreted by late morning or early afternoon. That’s because their physicians, in an effort to make it convenient for them, have set up both imaging and clinic appointments on the same day. That makes a lot of sense, but we need our work lists to communicate that, otherwise we have no way of knowing if one outpatient actually needs to get interpreted before some other outpatient. Smart work lists are absolutely critical.

[Communication is] something we spent so much of our time on. There are a lot of solutions out there, both electronic and otherwise. Establishing asynchronous communication with referring providers is usually about noncritical findings. If there’s an acute bleed or a similarly urgent finding, you pick up the phone and call, but for findings like pulmonary nodules and incidental masses in the liver, you could still spend quite a bit of time communicating something that is not going to impact the patient adversely in the next hour, day, week, potentially even month, but could affect them significantly down the road. There are a lot of creative solutions for electronic asynchronous communication with our referring providers and also some creative solutions from different health systems for something as simple as knowing the right person to call for an emergent finding.

Radiologists are being asked to get out of the reading room for more face-to-face meetings, yet also must contend with high volumes of interpretations. Can you talk about this tension between visibility and productivity?

TC: It comes down to two competing priorities. You want to do the right thing for the patient on the screen at that moment, but at the same time you are beholden to all the other patients on your list as well. I think that’s where a lot of informatics solutions certainly have an opportunity and a role

For the communication problem, I know sites that have solved it with informatics and I know sites that have solved it essentially with people power. The [people power solution] is you identify findings you have to communicate to providers and send that off to a group of individuals in your department whose job it is to reach out to doctors and get them on the phone so you can relay the findings. It works very well at places that have the resources to provide that service 24/7. It may not be perfect, but it works.

Electronic alternatives are to have a system that lets you either reach out by phone or by text message and continue to ping the person until you either get a response or if a certain time elapses…it gets bumped up to a human who must intervene. When it comes down to it, this [communication] has to be recognized as a necessary part of the care that we deliver, and the resources—whether human or financial—have to be put toward that at some point.

You’ve spoken a lot about business intelligence, and there’s a growing interest in analytics in radiology. What metrics are currently being overlooked?

TC: To figure out what you’re going to improve and how you’re going to improve it, you have to figure out what—for lack of a better word—your deficiencies are, whether at the practice level, the physician level, or the group level.

Radiology has been notoriously bad at metrics. You may have an extremely high turnaround time on a case because it was a complex case and you had to spend 20 minutes getting someone on the phone to talk about it…none of that gets reflected. That is the new challenge that we’re faced with. Not really, ‘How do I start to improve things?’ but ‘What are the right metrics?’ Ultimately, we want to get to a point where we have patient outcomes reflected. It may not be long-term patient outcomes; it could just be short-term. Did they get discharged from the ED or did they get admitted to the floor based on the radiology interpretation? Was their length of stay affected by the radiology interpretation? Was there an opportunity to intervene more quickly because of the radiology interpretation?

Something that Paul Chang [vice chair of radiology informatics at the University of Chicago] says that I think is extremely applicable is ‘Business analytics is not something you buy, it’s something you do.’ It comes down to knowing what data is available to you and perhaps identifying what data would be useful to you and figuring out how to get that data. Start to put together things that are very simple, like patient wait times in waiting rooms. Where are the bottlenecks when a patient comes for an imaging study? The first waiting area? The changing area? Are they spending too long on the scanner table? Just looking at the numbers, you sometimes don’t know where the problems are and you may focus on the wrong part of the process. Maybe patients are always late to check in for their appointments because there’s a bottleneck in the garage that people never thought about. It’s essentially operations management; you have to look at the entire pipeline and some of the components that might affect us may not even be in our department.