Who Owns Imaging? (A Play in Five Parts)
This is not a trivial question. As digital imaging has matured, the stakeholders have multiplied, and a number of models have emerged. But consider this: ultimately, whoever owns digital imaging is responsible for uptime. The better question might be: Who wants to own digital imaging? “If we really recognize that digital-based image management is important, then it has to run 24/7,” said Paul Chang, MD, University of Chicago. “Who is responsible for providing the service, for the workflow, and for answering the phone at 3 in the morning? The panelists took absolute positions (for dramatic purposes) on a handful of different models. Dramatis personae were: Manuel Brown, MD, Henry Ford Hospital, argued for radiology control; Benoit Desjardin, MD, PhD, University of Pennsylvania, argued that each clinical domain should control its own images; Steven Horii, MD, University of Pennsylvania, argued that IT should run imaging; Chang made a case for enterprise IT; and Matt Long, Philips Medical Systems, offered the vendor perspective on the optimal ownership model. Radiology Rules Brown made a compelling case for radiology ownership in sharing some of his war stories from the 2002 replacement of what he called a primordial PACS: “We have a large corporate IT department, but we didn’t have much support from corporate IT,” he recalled. Henry Ford is planning to slowly expand its PACS to include cardiology, non-radiology ultrasound, ophthalmology, pathology, endoscopy, and dermatology. It intends to own the cardiology servers and interfaces, but will not support other forms of images. In defense of the radiology ownership position, Brown raised the specter of turf by flashing images of a dogfight: “Who owns you? If you don’t do this right you won’t own yourself. That fellow, the cardiologist, will own you. Will you be deputy dog or prison dog?” Having all images in PACS ensures a high availability for all images, Brown posited. “Common platform? Wonderful,” he said. “And if there is one, it should be under radiology because we have the knowledge and experience.” Home Rule Benoit argued the individual domains ownership perspective: “Every department involved in imaging will own its own piece: purchase, maintenance, financial control, design, architecture, upgrade.” His argument was based on the fact that cardiology, for instance, is more complicated than radiology, with more data elements to integrate, such as hemodynamics, ECG. There are more than a few points of difference between radiology and cardiology PACS: • data may have to be kept longer because heart disease calls for lifetime continuity of care; • cardiology data, even if DICOM, can be incompatible with radiology DICOM; • while everybody wants access to imaging related to their area, cardiology imaging does not need to be widely distributed; • ordering and scheduling varies by department; • sets of images can be separated by hours or days: obese patients will do resting and stress on different days; thallium imaging requires follow up in 24 hours on defects in fill; • radiology is a service organization, while there may be a time lapse in cardiology between acquisition and interpretation (ah, the leisure of it all!) • in radiology, the main product is a report, in cardiology, the report is an afterthought, though the information is critical to patient care; • huge data set swill clog up the network and also lengthen migration efforts considerably; and • who wants to trust another department to police data integrity? Give It to the Geeks Horii made an all out pitch for IT control of radiology. “We are the hardware and software geeks. Images are just big files, and we know how to move, store, and retrieve big files. We know how to do it securely. Concerned about patient privacy? Horii directed doubters to witness the insurance and banking industries where information needs to go places and it can’t be intercepted.” You think IT can’t handle disaster recovery? “Who do you think built Cheyenne Mountain, with spring-suspended servers so they could tolerate a nuclear hit?” We are the networking experts, he said. We know what you need. “Even better, tell us what you need to move, and we’ll tell you what bandwidth you get.” In a crescendo of IT machismo, he called for a no-nonsense approach to radiology IT. “And none of this sissy, fourth-generation language crap,” he said, drawing applause from the IT contingent in the audience. A Kinder, Gentler IT Chang shared a bit of his recent history to put his argument for the enterprise IT model into context. When the radiologist and developer of the Stentor PACS (now Philips) left the University of Pittsburgh and went to the University of Chicago to become medical director of enterprise imaging, the radiology department was required to give up ownership of PACS (a privilege the chairman relinquished reluctantly and probably only because he was handing off the PACS to another radiologist). Because the help desk model of IT support is not enough, Chang advocates enterprise IT ownership under what he calls the affinity matrix model. This incorporates weekly meetings of affinity teams—for instance, IT network and infrastructure representatives, as well as IT experts in the radiology domain. “IT has no role at all in trying to define workflow, functional requirements, use cases, or how you do your business,” Chang told the audience. “But I don’t think radiology has anything to say about segmentation of secured networks, either, assuming you have competent IT. We are pretty good about networks.” When it comes to cardiology PACS, Chang tells the cardiologists to pick their own. “I tell cardiologists, you pick what you want, you know the workflow, you own that part.” The trade-off is that the cardiology PACS invariably includes a mini PACS archive, but Change will grant only a limited persistence cache. “Just enough storage to support workflow.” Success of Chang’s enterprise IT model hinges on formalizing service level agreements with every department in the hospital to define the anticipated level of support. “This all gets down to accountability.” Observations from the Trenches Matthew Long, Philips Medical Systems, has the vantage point of having witnessed multiple PACS environments. He divided the ownership issue into two components: evaluation and vendor selection; and implementation and ongoing support. “In our world, PACS are inherently IT systems, but the user is a clinician,” he said. “If a PACS was static, I think you could make the argument that it would stay in the clinical domain. But it doesn’t stay the same.” Changes in data types, users, and interfaces result in at least one upgrade a year. “Radiology does not control the infrastructure,” he said. “Because of the dynamic nature of PACS, if you do not control the infrastructure, you do not control the technical resources and are not focused on enterprise requirements.” On the other hand, IT does not have the appropriate clinical sensitivity or understanding of workflow, Long insisted. He advocated two options: 1. A dedicated informatics team that understands clinical issues and advanced technical skills and can bridge the gap between radiology and IT. 2. Informatics tied to radiology but part of IT. “Those are the two that we see work best from an ongoing maintenance perspective.