Duke University Health System Selects Epic-friendly VNA
When Duke University Health System (DUHS) in Durham, North Carolina, goes live with its full-blown Epic electronic health record (EHR) implementation in June 2013, another equally important transition will have happened in the background: the installation of a new vendor-neutral archive (VNA) that will eventually contain all 135 terabytes of image data currently housed in the department of radiology’s PACS. That’s not the endgame, though, according to Christopher J. Roth, MD, radiologist and vice chair of radiology for health IT and clinical informatics at DUHS. Roth also serves as the enterprise director of imaging-informatics strategy. The plan is for the archive to host almost every DICOM and non-DICOM image generated by all specialties, now and in the future. “Looking at it from the health-system perspective, the three-to-five–year vision is to get as many of the other clinical specialties as possible up where radiology has been for years now, to provide a high level of functionality and security around the images that they create, and to accommodate emerging needs in medical imaging,” he explains. With an Epic implementation imminent, increasing government health IT requirements, and an ambitious plan to accommodate the imaging IT needs of all specialties, Roth and the IT department at DUHS were looking for a highly flexible, nonproprietary, simple approach to enterprise image-archive needs.
Christopher J. Roth, MD“Many of the applications—and the whole spectrum of health IT, at this point—are moving in a direction of seamlessly integrated, enterprise-level solutions, and that is what this is intended to be.”
—Christopher J. Roth, MD
Duke University Health System


Living in an EHR World More than one factor drove Roth and DUHS in the direction of the VNA. To begin with, there is the new way that radiology is viewing PACS. “We are beginning to look at PACS as something to be deconstructed and rebuilt in best-of-breed component parts,” he says. Typically, PACS is being disassembled into three pieces: The archive and image-management components (including life-cycle management, security, storage and distribution); the viewer; and functions related to radiologists’ workflow, such as worklists, incorporated peer review, and critical-results review and reporting. Just as one might migrate all data off a desktop or laptop to be decommissioned, existing images in a PACS archive are best migrated to vendor-neutral storage in the event that an institution is considering a new PACS, Roth believes. Clinical imaging-informatics apps are developing rapidly, and there is a growing need for the interpreting physician to be able to use the tools of choice to do, for instance, 3D reconstructions, quantitative analysis, and functional imaging—using the best tool for the job. Radiologists and other physicians do not want their choices limited by the storage medium, Roth adds. Duke Medical CenterAside from diagnostic imaging, clinical images being generated at DUHS include patients’ headshots, operative and educational videos, and digital pathology. “There are a lot of different storage needs required of a health-care system that didn’t exist when PACS was being developed,” Roth remarks. “A few years from now, everyone in the health system will have needs for storage,” he says. “Many of the other clinical specialties also are performing their own imaging, and they are performing imaging well, at the point of care, to make immediate decisions. Because their solutions evolved locally for a given department or clinic, however, what they are doing (inconsistently, in many cases) is maintaining a broadly and easily accessible longitudinal record of images taken.” Roth continues, “In many cases, ophthalmology and dermatology departments are taking images, but they don’t have a great way of doing anything with them. They may be sitting in a shoebox somewhere; they may be sitting in someone’s desk drawer. They may be printed out on a piece of paper and put in someone’s chart, which doesn’t do anything in an EHR environment.” The clock is ticking when it comes to digitizing patient information, Roth says. “These are things that ultimately need to be changed in the fairly near future, and radiology has been doing this for more than a decade, easily,” he notes. He believes that radiology has a partial responsibility for sharing what it knows with other specialties, particularly concerning the operational aspects of managing and distributing images (as well as making them available to patients). Choosing a Vendor To choose a vendor for its VNA, DUHS appointed a four-person selection team (a cardiologist and two enterprise-IT people, in addition to Roth) to work alongside the department chair and the health-system CTO and CMIO. First, the group issued a standard request for information, followed by an exceptionally detailed request for proposal (RFP) featuring 154 questions. The RFP was designed to probe implementation and support services, licensing, system requirements, any limitations on viewing apps (whether for PACS or for universal viewers that can present any kind of image), administration needs, maintenance, fault tolerance, security, auditing, and performance. All four members of the team arrived at the same conclusion, choosing the VNA from TeraMedica. Roth says, “There were some very specific things that were required, and as we went through and looked at responses, the vendor we chose rose to the top.” Criteria that were not negotiable included experience integrating with the Epic EHR and the ability to manage the existing PACS infrastructure at DUHS. Flexibility was another must-have characteristic. “That is important because you never know what the system is going to require at any point in the future, or what a given group of users will require, from a clinical perspective—or a research perspective, for that matter,” Roth says. “You need to have a flexible system: The whole idea behind a vendor neutral archive is flexibility.” Another important requirement was that the infrastructure be completely devoid of proprietary code, to smooth transitions from one PACS to the next—and even to the next VNA. “If we want to change archives to somebody else’s fantastic mousetrap, we want to be able to go from archive 1.0 to 2.0 easily, without having to do a bunch of extra contracts and a bunch of extra negotiation,” he says. “We believe we found a very flexible and powerful system.” On top of the RFP, the DUHS team also looked at Integrating the Healthcare Enterprise (IHE) compliance, investigated how vendors achieved interoperability at the IHE’s Connectathon events, and talked to people at a few existing sites for each vendor under consideration. Roth observes that smaller providers (with less-robust IT support) might not share the DUHS requirements for flexibility and a nonproprietary storage environment. “There are a lot of governance decisions around having that flexibility,” Roth notes. “The proprietary stuff, for us, was a big problem: We just didn’t want to have to deal with that, at any point.” Non-DICOM Images and MU Another deal-breaker was the inability to cope with non-DICOM images, for two key reasons. First, “This is not just a radiology archive,” Roth reiterates. “When the dermatology service takes a visible-light photo of a mole, you want to preserve that in as pure a form as you can. If people want to take that image and do things to that photo—and people do that all the time—you want them to be able to do that.” Second, Roth believes, an enterprise archive must be prepared for the unknown. “We don’t know what we may store in this system yet because it’s new, and we haven’t deployed it everywhere,” he says. “If there is something outside of DICOM that comes up years down the line—or something really terrific, today, that is very pertinent to a specialty outside of radiology—we want to be able to accommodate that. Limiting everything to DICOM handcuffs you.” Initially, the VNA will contain images generated by the radiology department, soon to be followed by images from the cardiology department. “We are having conversations with pathology and endoscopy to start incorporating their images as well, which are not today captured and distributed in any systematic way,” Roth says. While DUHS has not yet attested to stage 1 meaningful use, Roth speculates that the TeraMedica VNA could be helpful in simplifying the stage 2 requirement of providing access to images and reports by offering a single point of integration with Epic. “Cardiology has a number of systems; radiology has a single system; and ophthalmology, dermatology, and pathology are all developing theirs,” Roth explains. “If you are going to enable an EHR to use images, you have to build multiple interfaces out to all of these viewers—all of these different PACS, so to speak—or you can have a single platform that seamlessly talks to your EHR, and you can have everything sit there.” Above all, the objective is to keep it simple. “There are plenty of things for us in IT here at DUHS to worry about; one thing we don’t want to worry about is making our lives more complicated than they have to be,” Roth says. Cheryl Proval is editor of Radinformatics.com.