Of the many steps that Ohio State University (OSU) Health System, Columbus, took to ensure a successful integration of its picture archiving and communications system (PACS), the most prudent was its decision to train clinical personnel for duty in the information services department, the unit that would have primary responsibility for the ambitious, enterprise-wide linking of medical informatics technologies.
“Fully 40% of my information services staff are people who started as nurses or radiology technologists,” Asif Ahmad, MS, MBA, CIO and chief technology officer (CTO), says. “I recruited these clinical people as part of our planning for PACS. Once it was settled that we were going to acquire PACS and subsequently integrate it, I did not go out and shop around for the most highly skilled technical experts I could find. Instead, I took a manager from radiology who had the aptitude to learn information technology and I put her in charge of the PACS division. She then absorbed an enormous amount of vendor-based training, which brought her up to speed. Then, she and I started recruiting really innovative radiology technologists to round out the team.”
Ahmad continues, “This was a very logical strategy. To tackle the issues posed by PACS integration effectively, the organization must have at least some members of the information services team who come to the table with an intimate understanding of the enterprise’s clinical operations. My sense is that a lot of CIOs and technology planners do not really understand their enterprises’ clinical operations because they have come up exclusively through the ranks as purely technical specialists, not as clinicians. CIO and CTO offices are not usually attuned to how nursing provides care or to how physicians provide care. A lot of times, the decisions of CIOs and CTOs are dictated by technology trendsby what is new and innovativerather than made with an eye toward how those trends fit the care process.”
It helped, too, that OSU took pains to start down the path to PACS integration by asking and answering basic questions about the nature of integration and what it was supposed to achieve.
“Integration,” Ahmad says, “is a term that we all tend to throw around casually without really having defined it. To me, it means an optimized model for communication involving and enhancing an existing process (or, any series of events that leads to the completion of a given task). A real-life example would be the physician who performs his or her rounds in the hospital. This physician makes a stop at a patient’s bedside and wants to look at two things: laboratory results and images.” He continues, “Without integration, two separate systems must be used to access the information and images. Access to those systems might be difficult, however; the access point for one might be at the nurses’ station, while for the other, it might be one floor above. Because this is an inefficient arrangement, it will consume many minutes of physician time to retrieve the desired information and images. At best, the physician will have less time to spend at the bedside, actually delivering patient care. At worst, human nature being what it is, there is a likelihood that the physician will not want to take the time required to look up the images and data and will, instead, attempt to deliver the bedside care without that information in hand. Either way, this is a prescription for reduced quality of care.”
Ahmad says, “Obviously, this problem is solved by having the PACS integrated with the other technologies that provide laboratory results and related information so that care can be delivered conveniently at the bedside. Now, the communication has been optimized. The result is an enhanced process that translates into efficient delivery of care because the physician has access right there at the bedside to all the information needed for that purpose.”
At least that is the result in theory. Ahmad believes that, to make reality match the theory, careful forethought must be given to integration’s effects on work flow.
“We were successful with our own integration efforts not because we have the best technology, which in fact we do, but because we did an extremely thorough job of mapping out the processes of how care is delivered, identifying how technology fits into that and then designing the integration and interfaces to match those two,” he says.
Indeed, the fruits of successful integration at OSU are abundant, but one of the most striking is the way that it has freed non-radiology physician’s time.
“Thanks to the changes in work flow made possible by PACS and integration with other systems, including order-entry, electronic signature, and automatic results notification, we are saving 7,000 referring-physician hours per year,” Ahmad says. “This mainly represents elimination of the need for physicians to walk down to the film library to locate films. With the savings in time, our physicians are able to focus on doing what they do best: delivering patient care.”
Ahmad adds, “Meanwhile, on the radiologist side, PACS integration has increased productivity to the point that we have seen at least a 20% increase in CT revenues even though our number of radiologists and technologists has remained the same. It formerly took up to 90 minutes per CT scanner per day for the technologists to sort, label and transport films. In our fully electronic and integrated environment, those tasks no longer are performed. They have been eliminated. The resultant time savings allow us to perform studies on three additional patients each day per scanner, and we have five CT scanners. This increase of up to 15 additional patients daily has not slowed down the radiologists, however, since their ability to do work has also? been streamlined. They can read more studies per day, which translates into more business growth.”
OSU Health System is located mainly in and around Columbus, but it serves a population of more than 1.5 million across a large geographic area. Medical programs at OSU have for 10 straight years ranked as being among the best in the United States by US News & World Report.1 OSU comprises 40 facilities: four major hospitals and 36 specialty outpatient sites.
Total annual radiology examinations average about 300,000, consuming more than 7 terabytes of PACS memory and storage space. PACS access alone at OSU is logged at 700 to 800 users a day, much of it via secure Internet connections from laptops and desktop computers.
“We incorporated a web browser 2 years ago, so physicians can now log in from their offices or homes using our virtual private network (which is designed to allow fast access even with a lot of traffic present),” Ahmad says.
Almost all imaging modalities are tied into PACS, with mammography being the lone exception. Ahmad says that an OSU claim to fame is a totally filmless operating suite.
“About the only time that we print film is on demand, when a patient needs to take the image outside our system,” he says. “In film costs alone, we are saving $1.2 million annually. With film purged from this environment, we no longer needed to maintain a film library, and that allowed us to decrease staffing by 10 full-time equivalents, another huge cost savings.”
It was in 1995 that OSU decided to start its journey toward a filmless environment.
“The first step was to develop a unified vision of clinical computerization,” Ahmad says. “We soon realized that this vision needed to center on an electronic medical record that would include patient histories, charts, orders, laboratory results, and images, with all of it integrated so that clinicians and others involved in the delivery of care would be able to access all of the information and images needed at the bedside from a single platform.”
Integration stood as a particularly daunting challenge owing to OSU’s pledge to steer clear of proprietary systems.
“We made a commitment not to be held captive by any one vendor because we knew, in surveying the state of the industry back in the mid-1990s, that no one vendor could provide a full integration solution. We were right in taking that stance because, to this day, that remains the situation,” Ahmad says. “We decided therefore to secure informatics systems that were best-of-breed and then integrate from there.”
Not until 1997 was consideration given to the acquisition of a PACS. According to Ahmad, vendors prior to that time lacked a sufficiently developed PACS offering to satisfy the image manipulation, distribution, and storage requirements of OSU. When the information services team completed its 1997 evaluation of the PACS market, it chose an Agfa system.
“We went with Agfa because its technology included open-platform architecture, and this permitted us to purchase our own desktop computers and load systems onto them, saving us quite a bit of money,” Ahmad says.
NOT FILMLESS AT FIRST
The initial line of attack once PACS had been installed called for establishing and building up the radiology department’s soft-copy reading capability. It was not seen as feasible at that juncture to do away with film, since costs for long-term electronic data storage were prohibitive. By 1999, however, both memory and storage costs declined sharply, making it economically inviting to shift from a blend of film and soft-copy reading to nearly total filmlessness.
At about the same time, other initiatives aimed at advancing enterprise-wide clinical computerization were launched. These included planning for physician order entry, results reporting, and electronic signature. To facilitate all of these, the institution invested heavily in wireless technology.
“At one point we had the largest deployment of wireless devices in a health care facility anywhere in North America,” Ahmad says. “We now have about 150 wireless laptops on gurneys that physicians take on round with them”
In 2000, the PACS was upgraded to Agfa’s IMPAX R4 product. A year later, integration between PACS and the radiology department’s radiology information system was achieved. Fluent linkages with the central patient database followed in short order.
Because OSU chose a best-of-breed approach, one of the biggest integration-related challenges involved making the Health Level 7 interfaces that OSU had developed work flaw
lessly across the various systems. It might have been a more daunting undertaking were it not for OSU demanding, from each vendor, proof in advance that the systems under consideration for acquisition would at least be able to connect to one another (even if imperfectly, at first) once installed. This, Ahmad says, required a high level of vendor-to-vendor cooperation. Some balked, however, at the prospect of having to work with companies that they deemed potential rivals.
“We were ultimately able to convince enough vendors to cooperate,” Ahmad recalls. “What did the trick was our pointing out the potentially huge advantages of achieving success with their products at a major academic site like OSU. This helped convince the vendors that, if they cooperated? and advanced our common goals as a team, this could emerge as an important showcase for them.”
Ahmad adds, “CIOs and institutions often make the mistake of failing to hold vendors accountable. When they define a broader information-technology initiative of which PACS is a component, they have to take a hard approach with the vendors, and the time to do that is when they write the request for proposal. In it, they have to insist that the vendor’s products interface with other specified systems in order to win the contract.”
Vendors with which OSU was not eager to do business were those who only claimed to offer open-architecture. “Some vendors still want to marry proprietary software to a particular brand of computer or web browser,” he says. “They say that I can have integration, but only if I buy my computers from them. I do not want that because it can add costs. For example, the computer that would cost me $500 if I bought it from a discount house is, instead, priced at perhaps $5,000 if I am forced to buy it from the vendor, whose justification for charging 10 times as much is that it has been loaded with proprietary software.”
|Kathy Tunstall(left), associate director for PACS, and Asif, CIO and CTO.
Ahmad continues, “That is why I want my computing environment to be as platform-independent as possible, so that I will be able to deploy an application on as many different brands of computer as I own. We told the vendors that we would buy our own hardware and they would supply us the software. By doing this, our PACS deployment and integration were accomplished for about half the cost that other academic medical centers have ended up paying.”
THE INTERFACE PLAN
Another challenge for OSU was development of the interface plan itself. The enterprise elected to proceed with this planning in two ways. One involved a committee approach where plans were developed by consensus.
“This was effective because it provided the necessary input to take into consideration the needs of all our many different users of the various systems that we would be integrating,” Ahmad says. “The consensus process, however, can involve a lot of hard work and an investment of lots of time. For example, it took months to develop the plans for just our physician-order-entry integration. To do this, we had 15 people sitting around a table, meeting twice a week for 2 hours at a time, to work out the details.”
The second approach to plan development found OSU entrusting key decisions to a few highly technically skilled individuals whose choices were based on their own understanding of what needed to be accomplished. Ahmad notes, “One often must go this route when dealing with extremely technically driven solutions that nobody apart from the real technical experts understands.”
Now that the worst is behind him, Ahmad has had an opportunity to reflect on the entire integration process and has decided that there are only one or two things that he might have done differently.
“If anything,” he says, “I think that we should have been bolder in mapping out a goal for PACS deployment and the subsequent integration of it. I say this because it turns out that there is far more demand for our system than we expected. The reason for this is that the system has worked far more successfully from the users’ perspective than they imagined.”
Caught by surprise in this way, OSU did not acquire technology sufficiently upgradable to accommodate the demand easily.
“We’re soon going to have to make some extra technology purchases we weren’t planning on,” Ahmad says. “In order not to be in this position today, we probably should have explored with our vendors some kind of lease arrangement for the components most vulnerable to becoming quickly outdated.”
The most important piece of advice that Ahmad can pass along on the topic of PACS integration is to customize as little as possible.
“Academic medical centers differ from one another, but many make the mistake of believing that, due to their uniqueness, only a totally customized approach to PACS integration will suffice,” he warns. “If they go that route, what happens is that they end up with code that is completely outdated in comparison to industry standards by the time that integration is achieved. When the vendor comes out with its next upgrade of the standard system, the completely customized site will not be able to make that new technology fit the environment.”
Rich Smith is a contributing writer for Decisions in Axis Imaging News.
- Best hospitals. Available at: http://www.us news.com/usnerws/nycu/health/hosptl/topjosp.htm. Accessed November 21, 2002.