|Matthew A. Barish, MD (foreground), and Darren Dworkin, Boston Medical Center.
The best-of-breed strategy for picture archiving and communications systems (PACS) is also called unbundling. It is a way of taking the most desired products from a given category-monitors from one manufacturer, for example, and scanners from another-and linking them together with software designed to create a functional, economical PACS. It is a low-cost alternative to purchasing a complete PACS (a turnkey installation) from a single vendor. A best-of-breed strategy can save money, but it is not without risk. The institution opting for it must bring some expertise to the table. Administrators must also be tough enough to impose control on the multiple vendors whose systems are being combined because those systems must work together.
One sizable institution that felt that it had little alternative to a best-of-breed approach was Boston Medical Center, Boston, Mass. The hospital began planning its PACS in 1995 and has had it up and running, with increasing sophistication, for about 2 years. Boston Medical Center, which now has about 600 licensed beds, is the child of a merger in 1996 between Boston City Hospital and Boston University Medical Center Hospital. Boston Medical Center is a not-for-profit institution that serves as the teaching hospital for the Boston University School of Medicine. It is also the largest level-I trauma center in New England. It sees many impoverished or uninsured patients. Recently, it has dispensed an estimated $166 million per year in free medical care. Boston Medical Center is a high-profile, patient-rich, cash-poor institution: a perfect candidate for a best-of-breed PACS.
The Sutter North Experience
Yuba City and Marysville, Calif, are sister cities separated by the Feather River as it runs south between Sutter and Yuba counties. The area is largely agricultural. Sutter North Medical Foundation (SNMF) operates outpatient sites in both cities, as well as serving the two counties. The distance between the SNMF medical office buildings in the twin cities is only 4 miles, but even that short distance was enough to create a black hole when it came to ferrying radiological films from one site to the other. SNMF’s CIO, Geoff Kauffman, says, “When you move thousands of pieces of film every week, film gets lost. We were looking for an electronic solution.” That solution turned out to be a PACS for which a software routing and connectivity installation by AMICAS, Inc, Newton, Mass, a health information technology company that specializes in creating web-based imaging systems, turned out to be the key component.
SNMF’s installation of its PACS was timed to take advantage of the foundation’s new diagnostic imaging center, which opened in the spring of 2001. SNMF has 72 primary care and subspecialty physicians on staff, but only two radiologists. Its modalities cover the normal spectrum of CT, MRI, ultrasound, nuclear medicine, and conventional radiography. The new PACS, which is just now becoming fully operational, lets the radiologists read at high-end gray-scale monitors enhanced with $15,000 video cards, Kauffman says. Clinicians have access to the same information using desktop computers with $500 video cards.
“We have a large server with dual processors with about 1 gigabyte of memory,” Kauffman says. “We have about 300 gigabytes of storage in our main server, and we’re adding a digital versatile disc jukebox for long-term archiving. Basically, all our modalities are connected to that system. When a study is completed, it is captured by the AMICAS system and then placed into the queues of different radiologists.”
The AMICAS software interfaces with a radiology information system package that creates an order for a specific patient from a referring physician. When a study has been read and its report has been transcribed, the radiologist signs it electronically. The images enter the PACS, but even before then, the images can be accessed by the SNMF physicians, if needed.
To develop its PACS, SNMF brought together its radiology chief, a radiology manager, and Kauffman. “I made the technology decisions. They made the functionality decisions,” Kauffman says. “AMICAS was far superior in price and one of the top choices for functionality.” The arrangement with AMICAS is a per-study fee, adjusted quarterly. Believing that it would generate about 20,000 studies per year, SNMF arranged to pay for 5,000 studies quarterly, with what Kauffman calls a truing up to cover discrepancies in volume taking place each quarter.
Kauffman says that the use of Web-based technology was an attractant when it came to choosing AMICAS. “Many software vendors were proprietary, cumbersome, huge, and incredibly costly,” says Kauffman. “With an Internet base and a software-only solution, you cannot get away with that cost.” He estimates that the AMICAS-driven PACS was installed for about one third of what a single-vendor PACS would have cost. “We have this thing so interconnected and interlaced; we have four vendors’ products in the system, so making sure that they all talk to one another is critical.”
SNMF’s PACS is new, but it is winning converts among clinicians. Kauffman says that a target of 2 months has been set for going filmless. The radiologists were quick converts. “In 2 weeks, they were reading 40% to 50% of studies on the monitors,” Kauffman says. He offers two caveats from his experience with SNMF’s PACS installation. First, make sure that modalities really are compliant with the Digital Imaging and Communications in Medicine 3.0 standard. Second, do not be afraid to confront the multiple vendors involved in the implementation. “They have to believe that you mean what you say totally,” Kauffman says.n
Matthew A. Barish, MD, is vice chair of radiology and section head for body imaging at Boston Medical Center. He is also an assistant professor of radiology at Boston University School of Medicine. Barish holds degrees in biomedical engineering and understands computers, and he was one of a self-described handful of Boston Medical Center radiologists who were early in pushing for a PACS. “My engineering background helped vastly,” Barish says. “It gave me more of an interest. I knew about pixels, rendering, and networking systems.”
Boston Medical Center has the equivalent of about 18 full-time radiologists. At its two campuses, it handles close to 250,000 imaging studies per year. Each of its adjacent campuses (which are the sites of the now-merged hospitals) has a typical array of modalities available, including CT, MRI, ultrasound, nuclear medicine, and conventional radiography. The two sites have, between them, according to Barish, 10 high-resolution two-dimensional workstations with Aurora Technology software, and four of those 10 are three-dimensional (3D) and enabled with Voxar software that integrates with Aurora. Three additional workstations are solely 3D, bringing the total of radiologists’ workstations to 13. In addition, there are about 6,500 desktop computers on the Boston Medical Center campuses and at a few allied outpatient centers. These computers have Web access to the PACS and allow clinicians, specialists, and conferring radiologists to receive electronic images.
The Boston Medical Center PACS is not complete. Still to come, Barish says, are digital radiography and computed radiography installations that will help Boston Medical Center lower its outlays for film. These additions will bring Boston Medical Center’s overall electronic image storage needs to about nine terabytes annually. “We did not go to any one dominant vendor,” Barish says. “We did it with a PACS consultant, the hospital information-technology department, and a very few interested radiologists and our chair.”
The Software Link
Kenneth Burgess calls himself a PACS planner. He is the PACS consultant at Boston Medical Center. He has been involved with the planning of the Boston Medical Center PACS since its inception. Burgess says that the heart of the PACS is the software developed by AMICAS, Inc, a Newton, Mass-based health care information technology company. The AMICAS software links together the various imaging modalities, the workstations, a radiology information system (RIS), and the storage archives. It is because of the AMICAS software that Boston Medical Center has been able to pursue its best-of-breed PACS strategy, Burgess adds; this is a strategy that has cut to less than half the estimated cost of a turnkey PACS installation. “AMICAS created a fee-per-study scenario where the cost of hardware was very low, compared with what it would cost to go out and buy a full archive,” Burgess says. “Boston Medical Center bought the hardware and AMICAS put in the software.”
Excluding the annual fee paid to AMICAS as an application service provider, the PACS installation, to date, has probably cost between $500,000 and $600,000, Burgess estimates. Upgrades and expansions, of course, have meant (and will mean) additional expenditures. Barish estimates that Boston Medical Center’s original outlay was about $250,000.
The Web-based AMICAS system was attractive to Boston Medical Center because it allowed the dissemination of images to clinicians using desktop computers and because the hardware was affordable and highly adaptable. “For a radiologist, any computer becomes a (secondary) workstation,” Barish says. “Wherever I am, I can answer a question about a specific case. Outside of radiology, the referring physicians use the system on a daily basis to look at cases and for teaching. It has eliminated teleradiology for us: that has become Web radiology or Internet radiology. I have been in Europe and have looked at our images from there. In 2000, we had a blizzard in Boston, and I could not get to work. Several of us looked at images over the Web, while those closer to the hospital could go there and read. That ability has been extremely useful to us.” In addition, Barish says, the future is likely to move even more in favor of Web-based technology, leaving Boston Medical Center in an adaptive position. “We will see the integration of the PACS much more in the hospital 5 years from now,” Barish says. He adds, “We will see the PACS in the operating rooms; they will have the high-definition monitors. PACS, and in 5 years, almost everybody will be looking at images through the Web.”
As Burgess explains, “When an image comes out of the MRI scanner, for example, it goes right to the AMICAS server farm, which is located in the hospital’s data center. The AMICAS system recognizes the image through a study identifier. It says, ‘This is not only an MRI, it is a neurology MRI,’ and it routes it to a specific workstation.” The AMICAS server farm, consisting of two Windows 2000 machines and a tape library, handle the DICOM import, database, Web server and archive capabilities. In addition, a status log feature, AMICAS Watch ™ keeps track of when and by whom a study was viewed and read. This eliminates duplicate interpretation by the Radiologist and adheres to HIPAA security. Within minutes of the patient completing the radiology procedure the images are available to the radiologist and clinicians.He emphasizes that the AMICAS software not only provides connectivity between modalities and other software packages, but also is designed-and must be designed-to expedite the continuing flow of information in the most usable and responsive way. He says, “How people work with the system becomes so critical in whether a PACS is going to be successful or not.” Burgess mentions patient validation as an example. Boston Medical Center sees many patients who are unnamed. They may not be able to provide a name, or they have no identification. If they come back a second time for treatment, how does the system spot a prior visit? “A medical record number is not enough,” Burgess says. “Does the system have the ability to search and query whatever minute piece of information that you might get to identify that patient? That is where working with these companies rather closely is so important. Are they willing to write the software for it?” Burgess says that identifying patients or handling annotation on screen without the destruction of the image are just two of hundreds of the kinds of process decisions that must be made in creating a PACS. For example, he notes, the combinations of AMICAS, the RIS, and the CT scanner that will be used must be decided. “Unfortunately, if you are not careful, your decision may not be supported by the system,” he says. In best-of-breed PACS, the roles of the various vendors are magnified by the fact that they must integrate seamlessly. When they do, the savings from use of the best-of-breed strategy can be significant in many areas.
Burgess says, for instance, that the AMICAS software solution that Boston Medical Center chose allows it to build its own workstations inexpensively. “We would actually sit down and build workstations. We would buy a commercially available desktop computer, load the software, put on dual monitors from another vendor, install the video card, and have a workstation for a fraction of the turnkey cost.” Even with all desirable features, a best-of-breed workstation would cost around $27,000 and a more stripped-down model would cost about $23,000, Burgess says, but a turnkey workstation might cost $80,000 or more from a traditional supplier.
Building its own workstations gave Boston Medical Center another advantage: its own staff could then repair and maintain them. Boston Medical Center has no money for workstation service contracts, which normally cost 5% to 11% of the workstation’s purchase cost, Burgess notes.
The IT Department’s Role
In the beginning, Barish says, the Boston Medical Center PACS was driven by the radiology department, but, as it has matured, a large role in the building, upgrading, and maintenance of the PACS has shifted to the facility’s information-technology department. If he had one thing to do over, Barish adds, it would be to bring that department on board from the start. “At first, we got by without either much help or much interference from it, depending on one’s viewpoint, but once the imaging starts to hit a certain volume, it is an absolute necessity to get the information-technology department involved. Recently, it has taken an extremely active role, and it has been extremely helpful in providing services and support, as well as in decision making for networking, access, providing computers, security, and all the things that really are key. Keeping all those systems running and working well is something that you cannot do without their support.”
Burgess seconds the idea that the radiology department must work with the information-technology department to make a PACS implementation successful. “Critical would be it, in one word,” Burgess says, “because 90% of the patient information is in the form of images. Those images belong to all those who treat the patient. The information-technology department creates the roadway for that to happen. In the case of Boston Medical Center, the AMICAS servers reside within the information-technology department. It has the personnel who watch over them and keep them running.”
Darren Dworkin is Boston Medical Center’s chief technology officer. He runs the information-technology department. “The information-technology department brought in the infrastructure design and provided the assurance that the technologies that we were bringing in would stay consistent with our hospital’s RIS,” Dworkin says. “We brought the RIS company and AMICAS together. We were a beta partner with AMICAS, and, really, we were instrumental in bringing about a lot of the design that they use. AMICAS is the big player in our central strategy.”
Dworkin reports that the best-of-breed plan does marry the institution to its vendors. “We need cooperation from AMICAS, and we have had that. The RIS provider’s cooperation also needed to be ensured. Some of this, we achieved through contract negotiation. We tied it to a commitment to work with us through integration. We cannot steer around the RIS company and AMICAS. For some of the smaller vendors, we said, ‘Listen, if you want to come to the table and be part of our strategy. this is what you need to do.’ As an institution,” Dworkin adds, “we believe strongly in the best-of-breed approach. What makes it work is the collaboration between vendors and the information-technology and radiology departments.”
What happens if the vendors drag their feet or, worse, pass the responsibility from one to the other so that the buck never stops anywhere? This is the risk of the best-of-breed strategy, and no institution contemplating a best-of-breed PACS should go in blindfolded when negotiating with potential vendors.
“The biggest downside,” Barish says, “is that, when something goes wrong, there can be a tendency for the various manufacturers to disown the responsibility for what has occurred. Things tend to occur between systems rather than within a system. Our most common problems have to do with connectivity.” Barish says, “It takes a strong presence, from the radiology department and from the facility, to say, ‘You both own this thing, and it is your job to fix it.'” Barish says that vendor conflicts have, however, been extremely rare (and, so far, dealt with quickly). “The benefits clearly outweigh any of those risks,” he notes.
Burgess says that Boston Medical Center hopes, at the least, to break even financially with its PACS. Those savings will come with greater efficiencies in image reading, reductions in film costs, elimination of lost film, and (perhaps) decreased liability payments, since there will be no lost film. Burgess says that Boston Medical Center uses about 93,000 m2 of film annually. Barish reports that virtually all ultrasound imaging is now filmless. He says, “We have had some difficulty shutting off CT and MRI films for specialists and in the operating room, and for outside transfers and second opinions. Very shortly, we are going to stop printing CT and MRI film completely, with the exception, perhaps, of some operating-room cases, mostly orthopedic.” It will be 2 years or so before Boston Medical Center goes entirely filmless, Barish adds, but he says that the radiologists have adapted quickly to reading from monitors. “We have become very dependent on them,” Barish says. “I find, from a physical standpoint, that it is difficult to read from film now. When viewing slices up and down, you can see changes in vessel or organ structure that it is difficult to see on film because, on film, your eye is moving from image to image, as opposed to the images staying in one place and moving under your eye.”
Burgess says that Boston Medical Center’s shoestring PACS budget may have worked to its advantage, forcing it to be selective and cautious in its vendor and system decisions. “It is very easy to overcapitalize a PACS,” he says. “The technology of PACS is overrated. You have to spend considerable time on how process flow can be assisted by the technology.” One way that AMICAS did this for Boston Medical Center was by creating middleware that translates Health Level 7 RIS data to the Digital Imaging and Communications in Medicine language of the PACS. “AMICAS is both the traffic cop and the translator for the storage connected to the RIS. AMICAS routes the information, stores it, and makes sure that you can pull it back as a prior study,” Burgess says.
Because a PACS installation is complex, particularly for the institution using a best-of-breed approach, Burgess urges heavy up-front planning and a go-slow attitude. “Plot your process flow,” he says. “All the sacred cows get exposed. Just by going through the process, you will be able to simplify it. It is very much like a value-added management process. Just in considering going to PACS, you are already doing something useful, even if you do not buy a thing.”
Burgess advises a three-step procedure for vendor selection consisting of a request for information, a request for proposal, and, finally, a price quote. “That is the way to filter vendors,” he says. “Spend the time planning. That is the most important thing to do, and it is the thing that people hate to do the most. Be cautious: there are an awful lot of PACS out there with which people are not totally satisfied. It is still an emerging technology; it is hardly mature.”
For an institution at a financial disadvantage (as Boston Medical Center’s health care commitments have forced it to be), Burgess sums up the planning process by saying, “Think things through. Thinking is still the cheapest thing to do.”
George Wiley is a contributing writer for Decisions in Axis Imaging News.