Gary Fammartino, MBA, (left) and Skip Beltz, MD, at St Vincent-Indy, where radiology will provide subspecialty reads for the network.

Good things do not always come in small packages. St Vincent Hospitals and Health Services (SVH) operates 11 hospitals and health centers across Indiana. They range from the flagship St Vincent-Indianapolis, which is licensed for about 700 beds, down to specialty clinics and rural hospitals that count their beds in the 20s.

Moreover, the SVH chain itself is a subsidiaryone of the most profitable and well-managed units, according to its administratorsin the giant Ascension Health group. Headquartered in St Louis, Ascension is the nation's largest Catholic health system and also the nation's largest nonprofit health system. It includes 67 acute care hospitals in 20 states and has more than 100,000 employees. So, for SVH in the first instance and with Ascension completing the bigger picture, there are serious incentives to adopt economies of scale.

SVH did this in a daring way a year ago when it built a storage archive designed to centralize the collection of all the electronic data from all the SVH hospitals into one mammoth repository. So far, only four of the SVH hospitals are fully on the system, but the others are scheduled to come on between now and 2006. When that is completed, radiologists at any of the SVH facilities will be able to go to their workstations and call up current and prior images from any other facility. Moreover, data from pathology, billing, and other pools will be on the archive, giving SVH what will amount to an enterprise-wide EMR (electronic medical record) for any patient in the system. If SVH succeeds with this central archiving effort, then similar efforts may be made throughout Ascension. Although it is not labeled as such, the SVH archive can be viewed as a pilot project.

The electronic health care universe SVH is creating is not being put together piecemeal. When St Vincent-Indy went filmless with its PACS (picture archiving and communications system) last year, it immediately stopped hard-copy filming for 65% of all imaging. SVH-Indy's Administrative Director for Medical Imaging Services, Gary Fammartino, MBA, calls this "the Big Bang approach." But there would have been no SVH Big Bang without some meticulous planning. Two of the key early decisions SVH made were particularly importantand not without risk.

CENTRAL ARCHIVE

Fammartino says administrators at St Vincent Hospital in Indianapolis had the original vision of creating a central system-wide archive at SVH. As far as Fammartino is aware, creating such an archive had not been attempted on this scale before. The idea for the archive was hatched over 2 years ago, before SVH-Indy had installed a PACS. One of the demands the chain made in selecting a PACS vendor was that the chosen PACS could be integrated with its planned central archive. Fammartino was working as a PACS consultant when he was hired by SVH to draw up an RFP (request for proposal) for a PACS. Originally, SVH considered six vendors, he says. "And we were able to pare that down from the responses to the questions on the RFP, which ran to 30 or 40 pages." After on-site demonstrations by two vendors, the SVH selection team narrowed its choice to one.

With the PACS vendor chosen, SVH began working to assemble its archive. Fammartino says the focus on a centralized archive was one reason the PACS vendor was not asked to supply the archive. Fammartino says a combination PACS/archive system was viewed as limiting. "We wanted an archive that could house not only our PACS but our EMR. We wanted an open architecture. We wanted to integrate into a SAN (storage area network) solution. Many of the PACS vendors had their proprietary solutions."

SVH hired a second vendor to help it build its open architecture archive. Essentially, open architecture means that hardware and software used in the application, in this case the archive, are equipment that can be interfaced using technologies that are bought separately or are in the public domain. According to Fammartino, the centralized archive that SVH built uses a cluster of servers and hard drives to store images and other data for immediate retrieval. Servers are computers that collect data and reroute it to other computers, along with instructions for its use. SVH now has about 10 months' worth of storage online for immediate retrieval, but this is being expanded to 18 months. For long-term storage and for backup, the archive uses high-density tapes.

"It's a network of servers tied together with [taped] redundancy," says Fammartino. "Not only does PACS go in there, but the clinical data also. So we have images, radiology reports, labs, nursing notesthe whole patient record is pretty much on that SAN." Fammartino says the radiological images come from the modalities into the SAN and then are automatically duplicated on tape. All the data from the various sources comes into the SAN and then is distributed back out of the SAN to the various users of data, including radiologists at their workstations. This collection and redistribution is accomplished using T-1 and DS-3 lines, which are high-speed telecommunications connections. The whole system is hard-wired. As each hospital is brought into the SAN, the hard wiring will be extended, mostly through DS-3 lines, a method of moving amounts of data instantaneously.

"The DS-3 line is not zinexpensive, but  because it carries other information too that is vital, like the financial, and there's the possibility of doing telemedicine," Fammartino says. "That's where we felt the robust investment was really worth the expense."

St Vincent-Indy, St Vincent Carmel, St Joseph Kokomo and the St Vincent Oncology Center are fully on the archive. So is St Vincent Children's Hospital, which is in Indianapolis. St Vincent-Carmel is in an Indianapolis suburb about eight miles north of the main hospital. Its proximity to the flagship was one reason it was brought on first. St Vincent-Jennings in North Vernon, which is about 60 miles south of Indianapolis, is partially connected but it comes into the archive on a separate server because it does not have a RIS (radiology information system), Fammartino says.

STANDARDIZATION

A second major decision taken by SVH as it formulated its archive strategy was to standardize modalities and other equipment throughout the SVH network. Fammartino says part of his job is to oversee this effort, which is occurring as new modalities and equipment are purchased. "It makes the archiving and the PACS easier, but it also gives us better pricing and it makes it easier for our technologists. We want to be able to share resources. We want to standardize our protocols so that we can extend care from Indy. We want the same protocols to go out on the system in the rural areasthe clinical protocols associated with, for instance, slice thickness for specific CT scans."

Fammartino admits there is risk in standardization, especially if the vendor or vendors chosen are late to develop new technology in the future. But he says the financial gains from bulk or cumulative sales offset that risk. "If there were a breakthrough in technology, we would consider switching," he adds.

By centralizing its archive and standardizing its modalities, equipment, and protocols, Fammartino says SVH will gain in efficiency and flexibility. Centralizing the archive gives SVH better control over the data than having it spread out on servers at all the various hospitals, Fammartino says. He also says the centralization allows a single group of archive technicians to tend to the storage equipment and troubleshoot any glitches.

GETTING THE BUGS OUT

There have been glitches. Rich Banta is senior enterprise systems engineer for SVH. He is a self-trained MSCE (Microsoft Certification Engineer) who was a chief technical officer for a medical network before coming to SVH.

"The archive designed by the vendors was pretty much nonfunctional, so we re-did it," Banta says. The big problem was that the initial design did not create enough storage space to handle all the data being sent to the archive, Banta says. "The tapes would hold only 18 gigabytes and 638 gigs was all a tape library would hold." The backup was writing everything to two tapes, Banta adds, but there was no repository for the second set of backup tapes. They had to be carried around in a cardboard box, Banta says. "Also, the first tape uses the second tape to heal itself, so we had to drag that one out of the cardboard box. The tapes were extremely fast, but they weren't working because they weren't dense enough. We had to pull the copy out of the box twice a day. It was ugly."

The solution came in the form of newly developed tapes. The new tapes were capable of holding 180 GB, making them 10 times denser than the old tapes. The new tapes did not solve all the problems. They created a new problem, in fact. The capacity was greater, but the retrieval time, because of the tape density, was too long. The deep archive recall was up to 68 seconds, an eternity in the electronic age, Banta says. "A minute is a long time. That was unacceptable; the doctors wouldn't wait. It caused problems with the PACS system too. There is a finite amount of time with the ultrasound retrieval that it would wait. It would flag it as irretrievable."

Just as denser tapes helped capacity, purchasing newly developed, dense, low-cost disks for the front-end or online storage boosted the amount of data that could be recalled instantly, Banta says. "Now, we just have to redo the disk cache, and we're done." The revamped disk cache will create enough disk space that deep retrieval will be far less frequent, Banta says. Currently, the SVH archive holds 10 months of data online for instantaneous retrieval. Banta and the other technicians are creating more online capacity. The plan is to boost the capacity to 18 months.

The SVH archive is complex (Figure 1). By using storage on spinning disks for immediate retrieval and tapes for deep retrieval and emergency redundancy, Banta says, the SVH archive should have the quickness and the storage space to meet future needs. "This year we will have four terabytes [on disks], and in 2004 we're shooting for 12 terabytes. This will give us 18 months minimum on spinning disks, so the likelihood of hitting anything that needs tape will be remote. The doctors say they can live with 18 months immediate retrieval that will be in milliseconds."

Figure 1. The St Vincent Hospitals and Health Services archive is intended to ultimately meet the needs of 15 hospitals and health centers.

Because SVH's creation of an enterprise archive on a large scale was groundbreaking, even the archive vendor engineers were not trained to handle some glitches that came up, Banta says. "We are on the bleeding edge, and it hurts us sometimes." To compensate for the inevitable problems and delays, SVH was able to negotiate "early adopter" status with the archive vendor and receive "deep, deep discounts," Banta adds.

Despite the problems it had to solve, SVH has done so well with its archive in the first year that it recently won a storage innovation award from an industry trade group. "We get lots of tours," Banta says, "pretty much from all around the Midwest. We had some people in from Texas last week." The interest in the archive is in-house too. Departments within the hospital now want their databases included for common access. "The neurology people want in on it now," Banta says. "Everybody wants into the act. If you can write to a Unix or a Windows disk drive, you can write to our archive."

As for expanding the archive, Banta says that should not be a problem if technology cooperates. "On the back end, it's built to scale to 122 terabytes, and all we have to do is buy more tapes. If everybody in St Vincent's came on, that would last us 4.5 years. By then, we think the tape density will double, and the disk density will double as well." Banta says the parameters eyed for storage are 21 years of images and data for pediatric patients, 12 years for mammography patients, and 5 years for other adult patients. "It's going to grow and grow and grow," he says.

Asked if it is not risky to put all its storage eggs in one electronic basket, so to speak, Banta agrees it could be. "We are more susceptible to the whimsy of the electronic gods, and we have been hurt by outages. The possibility is there to lose studies. But in the hard-copy world, 8%  is the accepted loss rate. We think we would be hard-pressed to lose that much. We think the archive is less of a risk."

THE MONEY LINE

Gary Fammartino recently made a presentation to the Suburban Health Organization (SHO), a group of seven mid-state Indiana hospitals that are not part of SVH. Fammartino was proposing that SVH act as the storage service provider for SHO, putting the SHO hospitals on the SVH archive. Whether SHO will accept the proposal remains to be seen, but the proposal illustrates the confidence SVH has in its system. It wants to market the storage on its system it has room to sell. The same sort of storage arrangement as proposed to SHO may be made with Northwest Radiology Network PC (NWR), the radiology group that provides almost all SVH coverage. NWR is now deciding on which vendor to use on a new PACS of its own. NWR is deeply tied in with SVH (see story, page 24), and NWR president and CEO Skip Beltz, MD, says the radiology group "will probably pay them for storage." The final decision depends on which PACS vendor the radiology group uses, Beltz adds. "But our images will be fed into the hospital system no matter what."

Before SVH committed to its PACS and archive, Fammartino had to show administrators the math. His number-crunching was convincing. It showed that SVH would move into the black on its initial PACS/archive investment prior to 5 years. "We saw that by eliminating film and a number of resources, such as clerical positions, we would see a positive ROI for Indy and Carmel. About 4 years out, it turned," he says.

Part of Fammartino's presentation to SHO was a detailing of that projected Indy/Carmel ROI:

• Immediate savings from going 65% filmless, about $500,000 the first year. Projected film savings over 5 years as the filmless percentage increased, $3.8 million.
• Eight clerical FTE reductions the first year and 15 FTE over 2 years, a savings of $250,000.

So far, these projections are being met, Fammartino says. Another area of significant savings has come from decreased per-patient time on modalities because technicians do not have to wrestle with film. "We have seen a 20% increase in patient throughput in CT, mostly for outpatients, which has meant well over $2 million gross to our budget," Fammartino says. X-ray throughput has deceased by 5% (CR was implemented pre-PACS; DR was added post-PACS) and MRI by 4%, he adds.

Beltz estimates that radiologist productivity is up about 20%, and will increase even further with the installation of voice recognition technology. For now, reports are dictated to a different dictaphone for each RIS, a cumbersome process. Because transcription lags, radiologists themselves are faxing crude, handwritten reports on emergency department cases, eg, "no fracture," or "no active disease."

These are all gains that soon will eclipse the PACS/archive's estimated $5 million cost for Indy and Carmel. As more SVH hospitals come on line, the savings to the SVH network are projected to accelerate.

Fammartino estimates that for a 200-bed hospital the cost of the PACS without the archive will be $1.5 to $2 million. The cost of archiving will add $60,000 to $100,000 for online studies and $80,000 to $200,000 for long-term storage, he estimates. But against these expenditures, he sees returns similar to those detailed above. Overall, he pegs the cost of an average study on film at $4 to $5, compared with $0.75-1.40 for soft copy. Based on 75,000 annual procedures, he projects the soft-copy savings to be about $244,000.

At St Vincent-Indy and St Vincent-Carmel, about 250,000 imaging examinations are conducted annually, Fammartino estimates. These are completed on 33 imaging devices and can be read on 22 workstations. As of now, SVH has 78,400 patients on its PACS/archive database; for these patients, there are 245,705 studies comprising 6,744,481 images stored, which take up about 3TB of disk and tape space, Fammartino says. He projects the storage cost in the first year at about 5 cents per megabyte, but in the second year, that drops to 4 cents.

Besides the ROI figures that can be set down more or less in black-and-white, there are many intangible or hard-to-detail benefits that cannot be marked on paper but are nonetheless true benefits of the PACS/archive, Fammartino says. These include image diagnosis by specialists for rural patients and increased information to referrers. "SV-Jennings is very rural, and we are remotely doing all their reads for them," Fammartino says. "They have eliminated film unless one of their subspecialists wants it. Having our neuroradiologists (and other imaging subspecialists) available to read for them has greatly enhanced health care in the Jennings community. We have seen our MR business increase in leaps and bounds in that market because of it." Radiologist Beltz gives another example of such a benefit: the ability to transmit and read images instantly. "We had two hospitals (SV-Indy and SV-Carmel) only eight miles apart. A patient would go to the emergency department in one and then for follow-up treatment at the other hospital. We would sit and wait for film for hours. Now we do all the emergency department and all the night reads from central command at the main hospital. How do you figure your savings in being able to provide quick care for the patient? That's hard to calculate, but there is one."

YET TO DO

SVH is far from finished with its PACS/archive installation. Most of the outlying hospitals still have to install a PACS and connect to the archive. There is a schedule for that through 2006. In August, SVH will take a related step by unveiling a Web distribution system that will allow clinicians at the hospital and referring doctors in their offices tosee images for their patients. "We have the dollars released to do that, about $150,000," Fammartino says. "We will have two Web servers that will pull directly from the SAN. Everything on the archive will be available, both current and long-term." Until then, referring doctors will continue to receive image copies on CDs.

SVH already has emergency department PACS workstations in its Indy-Carmel hospitals. It is exploring the use of portable monitors in its operating rooms. "Two-bank monitors that can be raised or tilted on carts," Fammartino says. When the Web access gets running, he estimates film production will decrease by another 20%. When the OR goes to electronic imaging, he estimates film will have been reduced by 90%.

Fammartino advises against leniency on film printing once a PACS/archive goes in. "We didn't want to live in both worlds," he says. "We went live on a Monday, and the weekend before that I had all the motorized viewers shipped out."

George Wiley is a contributing writer for Decisions in Imaging Economics.