PACS Roundtable, Part 2: The Ideal PACS-EMR Relationship

For many years, PACS lived in its own world, built to facilitate the transition from analog to digital imaging and developed for maximum radiology department productivity and workflow. The technology succeeded brilliantly, solving many workflow problems for users and supporting tremendous increases in radiologist productivity.

Now that electronic medical records (EMRs) are near ubiquitous in hospitals and physician office settings, integration between PACS and the EMR is imperative.  In the second part of a two-part roundtable discussion, we asked a panel of imaging IT experts to address what the ideal PACS–EMR relationship looks like and its role in an enterprise imaging strategy.

Roundtable Participants:

• Louis Lannum, Director of Enterprise Imaging, Cleveland Clinic
• Kevin McEnery, MD, Director of Innovation and Imaging Informatics and Professor of Radiology, University of Texas, MD Anderson Cancer Center
• Rasu Shrestha, MD, Chief Innovation Officer of University of Pittsburgh Medical Center (UPMC), Executive Vice President of UPMC Enterprises
• Chris C. Tomlinson, MBA, CRA, FAHRA, Senior Director of Radiology, Children’s Hospital of Philadelphia, and Executive Director of Radiology Associates of the Children’s Hospital of Philadelphia

What is the optimal relationship between the EMR and PACS, and how does it differ from what you currently have and what would it cost in dollars and resource to achieve, and are there any other data sources that would help radiology improve its service?

Tomlinson: The relationship shouldn’t be as it is today, just bringing up an image, but more as others talked about: How can we really mine that data? How can we have information that’s searchable and comparable? Not just “Hey, I’m seeing this patient in the clinic, I am in the EMR, and I click the link and it launches the PACS or enterprise viewer.” I think it needs to get a little bit more integrated than that.

Lou talked about bringing up multimedia, video images, light imaging. I think the enterprise viewers have to get a little better than they currently are. CHOP has invested in enterprise viewers since the beginning, and a lot of clinicians struggle a bit with them because an enterprise viewer and a PACS viewer are two different things.

A lot of people are used to a PACS viewer. A lot of people fight to keep the PACS viewer, because they want to see the radiology images, but we have to convince them that it’s not just radiology we’re serving anymore.

As a radiology director, I partner with my IT friends, telling clinicians, “I know you like that radiology viewer, but look at all the other stuff you can see in this enterprise viewer.” When you think about the journey, the enterprise viewers continue to get better. A lot of people think, “Oh, enterprise viewer equals radiology viewer dummied down.”

It’s not about how we put 10 million 3D reconstructions onto the enterprise viewer. It’s how we can bring up images in the context of that physician’s workflow. If that doctor’s up in the SICU and wants to see all the patients on his service and use the enterprise viewer for rounding. He doesn’t want to have to go back and forth between the EMR and the enterprise viewer. They want to see the patients in their bed order, so when they’re rounding, they can see the images very quickly and not have to do a search per patient.

Enterprise viewers still have a way to go in reflecting the enterprise workflow versus a radiologist workflow. I know we’re investing heavily into that with some of the newer players that are on the market. Again, a lot of the backend of this is how do the other data sources help improve radiology service?

What do you perceive as the optimal relationship between radiology and the EMR?

McEnery: I’ll try to make my answer brief, but I could speak for a long time on this. I am going to respond in terms of the radiologist, because I don’t think the EMR redefines PACS necessarily, but I do think it redefines the radiologist and the expectation for information they have available to them.

We had a self-developed EMR for 15 years that radiologists worked inside of at MD Anderson. We’ve now switched to another vendor, and we’re about six weeks into that transition.

The principles learned over the past 15 years continue to apply today. The radiologists want clinical information from the EMR. As for clinicians, some are content to allow the radiologist’s report to determine whether image review is needed, while others review on PACS every study ordered.

As an example, I sat down to lunch today with a colleague, and he mentioned that he’s now created a template that can instantly bring my impression into his report. He doesn’t have to look at the images anymore, unless there is a need based upon the findings noted in the radiologist report or the report differs from the clinical presentation of the patient.

I think there’s going to be pressure on radiologists to create information, and not necessarily only images. There will be pressure to make access to images more efficient, to provide the clinical context, so the radiologist can decompress the clinicians and allow them to provide better patient care.

However, the radiologist—as a quid pro quo—has to understand the patient’s clinical presentation. The EMR provides radiologists—as well as the referring clinican—with access to pertinent clinical information. It would be really great, too, if I could read the clinican’s note while they’re producing it, so that I could see what they were thinking and my report actually could be sympatico with the clinical presentation of the patient. Having radiologists and clinicians collaborating in the same EMR platform should provide all with better integration of all pertinent clinical information resulting in imaging reports that are informed by the entire clinical presentation and not simply by a single, provided clinical indication.  

My colleague agreed to do that. I think the relationship between EMR and PACS reflects how the relationship between radiologists and clinicians change, because they’re both working in the same environment.

Shrestha: Let’s piggyback off of what Kevin just mentioned. The way that I see it, we’re data rich and information poor. We’ve got lots and lots of data, but we’re really information poor. We radiologists, we’re fairly simple people:  We’re happy when we have access to relevant priors. Maybe we have some data from our 3D imaging system, maybe some data from the lung nodule tracking module. But there’s so much more data yet that exists in the EMR, and beyond.

The reality is that some of it may be relevant, and a lot of it actually may not be relevant in terms of what we’re doing specifically in radiology. The question really is how do we get at that data, but what’s even more interesting is how do we contextually represent that data, because context is king.

At Pittsburgh Medical Center (UPMC), we found when you’re looking across the value chain—it is critical to get at the data that might be relevant from the EMR that might help in ordering the study or acquiring the images (as we’re looking at protocols) or reading the case (and looking for specific nuggets of information that might be correlated with findings that we see in the imaging studies themselves)—if the context is king, how do we decipher the signal from the noise? Data itself is good, but what we seek is context so that we can get at the relevant insights that are, in turn, actionable.

The reality also is that not all of the data is in the EMR. To go back to the end of your question, are there other data sources? Not all of the data is in the EMR. As we’re looking at value-based imaging, we’re focusing on things like outcomes. If it’s outcomes that we’re after, how do we efficiently measure them? Some of it is in the EMR, and maybe we can correlate certain things.

We’re also looking at things like efficiency, and how we measure that. Also, we’re looking at lowering costs. There are other nuggets of data that the EMR might have bread crumbs and clues into, so we can really get at things like care coordination, referring physician and patient satisfaction and utilization management. It really is about connecting the dots across all of these different components of data that don’t necessarily exist in one silo. That really is the challenge—and the opportunity—with value-based imaging.

What are the benefits of building an effective integration between enterprise data and PACS? 

Lannum: At Cleveland Clinic, we found integration is about providing access to data—integrating all of these silos, either independently or bringing the silos into a central repository and making the data available to the physician, radiologists, referring physicians, orthopedic surgeons. It’s providing access to the data that they need, when they’re in the encounter with the patient, whether it’s a surgical encounter, or it’s an office visit.

For every imaging producer—whether they’re an orthopedic physician who’s taking an ultrasound or a gastroenterologist is doing a GI—there are 10 other physicians who want access to that data. Integrating that data into the EMR at the point of service is an essential part of our strategy.

McEnery: I think the point here is that the enterprise data can make PACS more effective, make radiology more effective. I’ll give you two use cases. One is in our switch to a new EMR, we’ve actually increased the fidelity of our orderables that the clinician can order. For example, instead of just ordering an MR of pelvis, we ask the clinician, what is the clinical question that’s being asked? Is it an MR of pelvis because the patient has ovarian cancer, or an MR of pelvis because the person has potential metastasis to the bone? The imaging protocols for these examinations are different, so it’s essential to have the study optimized for the patient’s clinical presentation.

Instead of just an MR pelvis, the indication’s pain, we’re asking the clinician to define for radiology and the radiologist what more specifically do they want out of the imaging that radiology’s going to perform? That’s one example. We’re using that information to make our protocols and our imaging more effective.

The imaging protocol also then informs the radiologist of what information they need to provide back to the clinician on their reports, again increasing the information back to the clinician.

Finally, in our worklist, one of the key fields is, “When is the patient’s next clinical interaction going to occur?” At MD Anderson, there is such a high percentage of patients getting imaged within 36 hours of their appointment—or even less—that we actually prioritize the worklist based upon the patient’s next appointment (not when the imaging was done), pulling the information from the hospital’s scheduling system. Again, we’re using information from the enterprise to improve the value, the data and the information that’s being provided from radiology back to the clinician.

Shrestha: Yes, I think we need to evolve the traditional view of what the VNA is to more of an enterprise content management system. Move , I think, a single-bucket strategy focused just in radiology to more of a capture-once, store-once and-use-multiple-times-via-the-right-means-across-the-enterprise approach.