The foundation of a fully electronic radiology environment is not the picture archiving and communications system (PACS). It is the RIS-the radiology information system.

Too often, this all-important piece of technology receives short shrift in the conversion to filmless imaging. Radiology departments and imaging centers sometimes are not even sure how to economically justify the acquisition of the ideal RIS package.

“RIS can make or break your hopes of being filmless and achieving all the things that go along with operating in that kind of environment,” says Toni Dudley, BSRT, computer analyst at Deaconess Hospital in Oklahoma City.

According to Michael Battin, project manager in the Information Systems Department at Hoag Memorial Hospital Presbyterian, Newport Beach, Calif, RIS-done correctly-yields dramatic cost savings.

“RIS can be seen as a well-justified investment,” Battin suggests. “However, there’s really no way to model the cost-benefit of RIS when you try to put hard numbers on it-it should be viewed simply as a required piece of equipment, an integral component for success today. Because of the functionality it provides, you can’t put a dollar value on it and say it’s worth only x-amount. It’s a self-justifying acquisition.”

VITAL FUNCTIONS

A good RIS must be capable of addressing several essential needs in the radiology department or imaging center, experts insist.

“RIS has to provide tracking of patients, films, and supplies,” says Billie Snow, information technical services clinical project consultant at Deaconess Hospital. “Our RIS, for example, includes a patient monitor that works like an airport traffic monitor. If a patient is sitting too long in the waiting area, the system will alert us so that we can address the problem. This monitor can be set up to flash the name of the patient in bright red to call our attention to the problem.”

For George Bowers, MBA, vice president and CIO of American Radiology Services in Baltimore, the single most important function of RIS is the management of work flow.

“PACS alone cannot address work flow,” says Bowers, whose 80-radiologist imaging corporation produces and interprets approximately 1.2 million studies annually at 18 outpatient centers. “Basically, RIS manages work flow from the point where the study is scheduled, to the point when the patient arrives, to the point where the examination is taken, to the point where the examination is read by a radiologist, to the point where the report is generated, and to the point where the film goes into storage.

“In contrast, a stand-alone PACS only manages images, gets them to the radiologist, and then stores them someplace-usually someplace independent of the report the radiologist creates when he reads the image.”

Ken Wright, RT, who works with Battin as Hoag Memorial’s PACS/RIS project analyst, concurs in this assessment.

“RIS should enable you to really understand where the bottlenecks are in the operations of the department, as well as provide some feedback that lets you know whether your attempts to fix them are working,” he says. “For instance, if we see 10 patients being registered within 1 hour for CT and the final examination is not being completed for 4 hours, scheduling needs to be examined. Additionally, a room’s throughput and patient wait times can be easily monitored via the RIS’ reporting functions; we use this in justification for additional equipment purchases.”

Wright adds that another key capability of RIS is report distribution, the mechanism by which radiologists’ findings are disseminated within the radiology department or imaging center and outward to referring physicians beyond those four walls.

Bowers’ system offers several options. “The RIS has the capability to distribute reports to the referring physicians in one of three ways: a paper report mailed to the physician’s office; a report automatically faxed to the physician’s office; or email,” he explains. “Email is very new and very few physicians are requesting their reports this way. The most popular way is faxing. The RIS usually requires that you build tables for the referring physicians with key information such as addresses, telephone numbers, and fax numbers. In addition, you can enter preferences into the table.; Some physicians like to have reports faxed to them before they have been verified by the radiologist. Others want only verified reports. The RIS is set up to trigger a fax the moment the radiologist completes a transaction, such as completion of transcription or final verification. The system also has the capability of enabling someone to quickly fax a requested report to a physician’s office. We use this feature a lot. In the future, we plan to make reports available online through a Web browser. We are beta testing software to do this right now. The limiting factor is the number of referring physicians with Internet access. Most of them have fax machines so that tends to be the preferred method of report distribution.”

In the list of important RIS functions, Bowers includes patient scheduling, patient registration, and order entry.

“In most hospital settings, RIS does not need to handle these particular tasks because they will be accomplished by the hospital information system (HIS),” he explains. “It gets tricky, though, in a HIS environment if you have got a RIS with a scheduling function. For example, you may want to use your RIS to schedule an inpatient for an MRI study. But if that patient has another procedure scheduled elsewhere in the hospital on the same day and at about the same time as the MRI, the RIS usually has no way to check for, let alone resolve, the schedule conflict. That’s why they usually leave scheduling as a function of HIS. It’s the easiest way to make sure a patient isn’t expected in two places at the same time.”

Not Optional

That being the case, some hospital-based department administrators are left wondering whether it might not be more economical to entirely forget about installing a RIS and just conduct business by piggybacking to the hospital’s information system.

“Such an approach clearly would be very economical-but when we gave consideration to this very issue 4 years ago, we concluded that the functionality of a HIS trying to cover radiology would not offer functions robust enough or radiology-specific enough to be able to manage the needs of our department,” says Wright. “That’s why we chose not to go that route and to instead implement a RIS.”

Some experts opine that in order to address the various needs presented to it, a RIS must possess certain hardware, software, and communications features. One is an architecture that supports system scalability.

“In an outpatient environment, scalability is extremely important, particularly if you are planning on opening more imaging centers,” says Bowers.

Central, too, to success is the RIS’s installed network bandwidth. An inadequate amount of bandwidth can make the RIS perform poorly, experts warn.

“Think of household plumbing when you consider the matter of bandwidth,” advises Bowers. “A pipe carries water. If the pipe is too small for the volume of water being pushed through, you have got problems. It is the same with bandwidth. It is basically a pipe that carries information. If the pipe is narrow, the information flow takes longer and applications do not run properly. You therefore have to make sure you have the correct bandwidth to serve up the information. But because bandwidth is expensive, you don’t want to oversize. Too much bandwidth wastes money. However, determining the right amount of bandwidth is not always easy, especially if your network will be carrying both data and images as in a PACS-RIS integration.”

For Bowers’ installation a TCP/IP Ethernet using T-1 frame-relay links the offices and hospitals covered. A single T-1 can move approximately 1.5 megabits per second of data. “At this point, we are carefully monitoring network utilization,” he reports. “We are not sure whether our existing network will have enough bandwidth when the full load of images as well as RIS traffic is implemented. Because we have outsourced our network and are using frame-relay, we have the capability of adding additional bandwidth without a lot of trouble. We use sophisticated network management tools to measure the utilization of our network at peak periods. We also have our network traffic segmented between the RIS and PACS traffic with the higher priority being given to the RIS traffic. Because the packet size with the RIS is so much smaller than the PACS, it has negligible impact on the response time of the PACS and yet ensures that the RIS performance is maximized.”

Hoag Memorial avoided that conundrum by installing one network to carry data and one to carry images.

“The regular hospital data do not commingle with the images, but they run parallel to one another,” says Wright, who notes that in this way the department is assured of having adequate, cost-efficient bandwidth for both its RIS and PACS needs.

The parallel networks are physically separate in topology but data can be switched at the bridge if needed, according to Battin. “The data network is a legacy Ethernet 10/100 MBPS all copper system,” Battin explains. “The new image network is dual gigabit Ethernet, with dual redundancy, effectively yielding 2 GBPS fully redundant. This translates to significant capital investment. An additional note is that we are replacing an older, less reliable ATM network with the gigabit system.”

As for hardware and software, American Radiology Services, for example, chose a product that operated in a Windows platform in order to gain access to the widest range of software programs and solutions.

“Windows is probably the most commonly used operating system,” says Bowers. “As a result, there is more off-the-shelf software designed for use on Windows systems than any other type of operating system. It is much more affordable to use off-the-shelf software than to have to develop it.”

CONNECTIVITY AT ISSUE

Health care organizations with distributed imaging requirements are likely to prefer an operating system that supports a thin-client environment.

“Thin client is a new thing and it is not perfected, but it is the way of the future,” Wright believes. “We like the thin-client approach because it requires so little management. In this organization we have more than 1,200 personal computers linked to our RIS and HIS. Deploying new software releases in a fat-client environment is just too labor-intensive for us. With thin-client, you install the software on your central server, and all the users whom you want to have it automatically get access from their terminals.”

Battin recommends using thin-client technology so that access to system controls can be tightly regulated from the server.

“You do not want people from their terminals going in without authorization and changing settings and reconfiguring the way parts of the system operate, as is possible in a fat-client environment,” he says.

Thin-client or fat-client environments notwithstanding, Wright holds that RIS hardware and software configurations should support a tool set that allows integration at the desktop level into either or both PACS and an electronic medical record system, and allow users to easily move between systems and applications.

“Many institutions are going to have information systems and imaging modalities from different vendors using different standards and protocols, and they’re going to need to link them together,” says Bowers. “Your RIS can be the linkage.”

Often, the requisite connectivity is achieved through interface engines that convert information from one standard to another. However, Wright argues that a better approach entails use of onboard brokers in conjunction with interface engines.

“The most innovative RIS products are delivered with brokers onboard,” he says. “That is very advantageous because it eliminates the need to rely on the latency of an interface engine and an HL-7 message sender. With broker technology, the modalities can actually get their work lists directly from the database of the RIS, almost on a real-time basis.”

Of course, the most basic means of ensuring connectivity is to purchase only those information systems and modalities that adhere to the HL-7 standard (and DICOM for anything PACS-related) from the outset, says Wright.

“If your RIS itself does not support HL-7, you are wasting your money,” he warns. “One of the things we struggle with is the reliability of interface connections and the performance of interface engines. We are solving that by decreeing that the particular set of HL-7 protocols that our HIS supports is the standard throughout this organization. So, during contract negotiations with vendors of RIS and modalities, we provide the HL-7 specifications documentation from our HIS vendor. This is what all our vendors have to contractually agree to meet.

“This leaves us with one flavor of HL-7 throughout our entire organization. This has been extremely helpful because our interface group now knows exactly what they are dealing with from every single vendor.”

Battin confides that taking this hard-line approach catches many vendors by surprise. Even so, they usually acquiesce.

“We insist on this way of dealing with vendors because we just want to create a conformant environment, which will hopefully allow us to reduce the amount of time we spend doing interfacing and also reduce our costs by allowing us to the extent possible to use off-the-shelf interfaces that will be compliant with other devices, information systems, and/or modalities via a standardized specification,” Battin explains.

LESSONS LEARNED

Perhaps the most important consideration when implementing RIS, experts say, is to bear in mind that there is no such thing as an easy RIS implementation.

Take, for instance, the experience of Deaconess Hospital where Dudley and Snow work. Deaconess is a 313-bed community hospital operating from a single location-it is 100 years old and one of the state’s last remaining independent hospitals. Deaconess recently completed construction of new quarters for its radiology department, placing that section on the first floor in the front of the hospital. This positioning is part of an effort to redesign the hospital using a medical mall format to make it easier for patients to get to the most frequently used services, according to Snow. Until RIS arrived 2 years ago at Deaconess, the radiology department managed work flow, scheduled patients, and monitored supplies entirely by hand.

“We went from the Stone Age to the future overnight,” says Snow. “In retrospect, I think it would have been helpful for us to have first gained a clearer understanding of the key problems RIS needed to address. We found that each RIS vendor’s offering addressed certain of these problems better than did others. Some vendors’ products did not really address them at all. Some vendors attempted to overwhelm us with their systems’ capabilities, which masked the fact that few if any of those capabilities would actually solve our problems or meet our needs.

“And, in retrospect, I think we perhaps should have prolonged our go-live date. We went live 9 months after we started building the system. Although 9 months seems like a long time, we probably should have taken an extra month for more testing and ironing out the bugs.”

Colleague Dudley offers that the toughest aspect of Deaconess’ RIS implementation was getting radiologists, techs, and clerical staff to change their work habits.

“Anytime you put in any kind of computer system, there is resistance to change,” she says. “If you’re going to get the maximum benefit out of the system, you have to change how people do things. To help our people go from doing things the old way-manually-we made up a very easy-to-follow manual for using the RIS. The manual was placed beside every terminal. This made it easy for users to understand how to do things that they may have forgotten since being trained. If they ran into trouble, they could quickly get the answers from the manual. No need to make phone calls and be embarrassed that they did not know how to do things.

“We also implemented bar coding, so there were fewer keyboard entry steps required to begin with to input information to the system.”

“The biggest change in our work flow is the interaction the technicians and physicians are having with the PC,” Snow reports. “It seems to have placed more responsibility on both as opposed to our clerical staff. But again, we came from a totally manual system.”

At Hoag Memorial, a 400-bed, not-for profit, church-oriented facility that logs 220,000 radiology procedures annually on-campus and at four off-campus, full-service imaging centers, the big lesson learned in implementing RIS was to create a full-time, dedicated team of radiologists, techs, and administrators to configure, install, and test the system.

“The biggest mistake a radiology department can make is to turn over sole responsibility for that to the information services department,” says Wright. “If you do, the implementation will not go smoothly. The information services department may not have the time or the inclination to lend as much assistance as would be necessary.

“So, radiology has to play an active role in the process. And the people you select cannot be performing a dual role-you cannot have someone assigned to this who is implementing one day and going back to being an x-ray technician the next, then back again to implementing the day after that-the process is time-consuming and difficult, so the people involved need to be immersed in it so they understand it from beginning to end.”

Importantly, Hoag treated its RIS as what it is-a mission-critical system. The implementation strategy that emerged took that fact into account, says Battin.

“We knew our RIS would undergo downtime at scheduled intervals for backup and other maintenance,” he says. “To allow for this, we put in place a backup transcription system. We switch over to this system during downtime. When the RIS is back online, we can then merge the information from the backup system into the RIS with relatively little human intervention.

“We also have the same arrangement set up for film tracking. A couple of hours each day, the database for the RIS dumps off to another system. Through a thin-client application, we allow users in a downtime environment to look up the locations of films and patients. That is not a product offering of any RIS vendor, but it is a mechanism worth looking into to help an organization be sustained through downtime and to help recover.”

Which is exactly what a good foundation should do in a filmless environment.

Rich Smith is a contributing writer for Decisions in Axis Imaging News.