HIStalk Interviews Jim Prekop, CEO, TeraMedica

Jim Prekop is president and CEO of TeraMedica of Milwaukee, WI.

Tell me about yourself and the company.

I’ve been in health IT for about 30 years. The last 10 have been with TeraMedica. Before that, I was in the EMR space and companies like PeopleSoft and Dun & Bradstreet software.

TeraMedica is middleware. The industry term is vendor-neutral archive. We collect clinical objects and are responsible for making them available to the source system, but also making them available in a patient-centric view to additional consumers of that data, whether they’re outside in institutions, exchanges, or new technology that gets adopted by the provider. We perform that role in the healthcare architecture.

How has the unbundling of PACS from single-solution vendors changed the demand for vendor-neutral archives and what’s the end result for the provider and the patient?

It’s a natural progression. With systems, historically, the new idea is a more or less a closed-loop answer. It’s the same way with accounting systems going back decades. 

What was a box has now become a layer in the architecture, the process of acquiring and managing an image and then making it available down the road to new consumers or later in my lifetime. The solution has had to evolve. The VNA, or the ability to seamlessly have the interaction with departmental activity but yet be the conduit into the enterprise, it’s a natural progression. It’s not to say that PACS is bad, just that the focus going forward on PACS will be different, just as the responsibility for the VNA will change over time as well.

What about universal viewers?

The universal viewer is interesting. They’re approaching this through the lens of the physician, whereas the VNA approaches it from the infrastructure up. 

The advantage for the enterprise viewer is that they can combine data from multiple sources. But the other thing that has to be kept in mind is that there is response time and there is certainty that is needed in what is delivered to the enterprise viewer. You get into a federated discussion of going after 20 different data sources, combining that answer, and then delivering it in one view to the clinician versus the ability to have all of that patient matching resolved by the VNA. It’s one-stop shopping. It goes to any consumer of the VNA.

We see the consumers being an EMR. We see the consumers being an enterprise viewer. Going forward as more adoption comes into the United States, it will be different exchanges that imaging will become part of. So to us, it’s just a consumer. We optimize its ability to be confidently assured that they’ve asked for and gotten the right information and that all the information is there. If you have a federated view and make a request and one of those systems is down, you might not get the answer.

Enterprise viewer implies that there’s behind the scenes fetching going on that then presents a unified view, as opposed to the VNA where it’s actually stored in a single system.

Yes. It’s already stored and normalized and you’re having one conversation behind the scenes. 

Unless somebody’s invented something new in IT that I haven’t seen, you pretty much have to ask the same question across multiple systems or go to some sort of index and find out all the Jim Prekops and then go and find out where they’re located, go get them, and then present it to me in an organized way. Can those enterprise viewers do that? Absolutely, and we have great partners in that space. Is it the best experience for the provider or the clinician? Maybe not.

What are the optimal ways to integrate a variety of images into Epic or Cerner?

I call it a landing page. EMRs address all the departments in the organization and rightfully so. But if I want to go look at all the different clinical objects that Jim Prekop created in a facility, chances are the links to that information are within various locations within the EMR. 

One of the advantages that TeraMedica brings to the table to leverage the investment that the provider has in the EMR is to give a patient-centered view of all the clinical objects, should they want that. That’s an option in our system. We can be tied to a report and just show that image, or we can present a complete inventory of what we have in the VNA, so that in one location, a clinician can see things that might be related to other departments. I don’t necessarily have to navigate over to that section of the EMR to see those objects.

It’s probably important to note that all images are objects but not all objects are images. Are you seeing demands for new object types?

Absolutely. When I first got here, I had to get an education on DICOM and all the nuances and it was a big education. But not everything is DICOM when it comes to clinical objects. 

Our customers asked us very early to not just manage DICOM. It’s a wonderful thing and is the heavy lifting in our business. But to be truly patient-centric, you have to address all different types of file types, whether it be JPEGs, MPEGs, PDFs, a Word document, or in the case of cancer care, lots of calculations are done using Excel and other types of planning systems.

To represent that an image is just a DICOM object is not fair. It’s usually one of the arguments when you try and decide what a VNA really is. There are lots of folks that manage DICOM and they do a good job, but they declare themselves as the VNA. That doesn’t meet our definition of a VNA.

What’s the distinction between storing non-DICOM data in its native format instead of using a DICOM wrapper?

Unlike other industries where you can create data marts and if there’s a problem you just snap another copy of the data, we’re into terabytes and hundreds of terabytes of data. As you acquire that information as the VNA, you have to be clinically responsible to the source system. If I go get a PDF of Jim Prekop from a clinical system and I wrap it in DICOM and that system wants it back, I either have to create duplicate storage — which is not cost productive — or I have to be able to unwrap it from that DICOM and enter that as a PDF to that source system.

The overhead of doing that simply doesn’t work and it doesn’t scale. To believe that you have to wrap everything in DICOM so it follows how your system works … I would suggest you have the wrong system if it only works with DICOM.

A well-known VNA consultant who comes from a PACS mentality is adamant that everything should be wrapped in DICOM. We needed to get him to sign an updated non-disclosure agreement, so I had my engineers wrap our NDA in DICOM before I sent it to him. His asked me what I had sent him since he operates on a Macintosh that doesn’t understand the file type, which is a .UCM. He didn’t even recognize that I had sent him a DICOM file. He didn’t understand that he was essentially justifying the reason why we believe that it’s DICOM and non-DICOM.

Who are your main competitors and how do you differentiate your product from theirs?

Since the VNA term was adopted — I prefer Vendor-Neutral Architecture — lots of folks put their hat into the game. As you would expect, a lot of PACS vendors have begun to open up and allow multiple DICOM systems to enter data in there.

It’s usually TeraMedica and Acuo that end up being the finalists in any evaluation. There are some other ones that are out there that do some of the things that we do. There’s some newcomers — Mach7 is out there, but I think they have more activity outside the US than they do within the US. But there are others that are coming into the space, and rightfully so. It’s a competitive market.

Hospitals acquiring medical practices and each other have left them trying to figure out how to get their systems to talk to each other. Is that true of imaging systems or other systems that would populate a VNA?

There’s two aspects of that. We’re having organizations that are buying us because they’re strategically positioning themselves to acquire other entities. They know that they can’t rip out those clinical systems, so they will use us as part of their strategy to get control of the data and share it across the enterprise.

As far as the other way, we have sites that are established either because of acquisitions or because of differences on campuses that have multiple EMRs. Our technology allows, again using myself as the example, Jim Prekop to be referenced, and if I know the request is coming from Epic, I’ll behave one way to put it properly in Epic. At the same time, I can put it into Cerner. There’s one source of the truth.

One of the value propositions that we bring as a VNA is that we can identify consumers and react accordingly. We can also respond to multiple consumers, but yet give them the exact data that they’re looking at, whether they come in through the physician’s office with one EMR or they come in through the hospital with another EMR. It’s one source of the truth with multiple consumers.

Where do you see the company going in the next three to five years? 

I think it’s based around being a good partner with our customers and bringing to them more use cases, more managing the data. As you would expect, we can sit behind a PACS, but the thing about VNAs is we’ve had to come around the curtain. We’ve always considered doing the plumbing behind the scenes. But now we’re very active in different departmental workflows.

We’re getting involved with our iPad app, as an example, in departments like wound care and dermatology, where the clinicians are actually interacting with our software and we are part of the EMR, but the clinician doesn’t even know we’re there. A lot of times when someone says, “I didn’t know you were there,” that’s a bad thing. For us, that’s a good thing, because we want seamless integration into these different systems. I can see us doing more of it.

I can see us taking responsibilities for more functions of a generic nature in the provider space so that they can optimize the platform that they’ve invested in. Clearly the leading investment is the EMR. But the VNA is also a strategic investment, and we need to do more for them when it comes to clinical workflow.