Computerization of healthcare solves a lot of problems, doesn’t it? It can relieve staff shortages, eliminate film and paper, enable faster patient throughput and data processing, and provide economies of scale previously only dreamed of. Conveniences like Internet patient scheduling, digital imaging and voice recognition save everyone time and steps, not to mention money. It all adds up to bottom-line savings, better workflow, and improved patient care. Now, if only all that computerization could be crammed into one big, do-it-all system — wouldn’t that be perfect?

It certainly would. And that’s exactly the problem resulting from all these wonderful, modern conveniences: so many solutions, so many vendors. How the heck do we integrate them all?

The megasystem concept
As it turns out, healthcare has been wondering exactly that while waiting patiently for the hospital system of tomorrow. Presumably it will be a many-splendored production, incorporating myriad functions previously processed independently on disparate servers.

Like the corporate mergers that swallowed mom-and-pop businesses everywhere, the hospital megasystem, if you will, will consolidate a facility’s departmental computers. Bit by bit, that’s already happening. The 2003 Leadership Survey by the Healthcare and Information Management Systems Society (HIMSS of Chicago) found that an overwhelming IT priority right now is systems integration in a multivendor environment.

According to the business research firm Frost and Sullivan (San Jose, Calif.), convergence is the word, with picture archiving and communications systems (PACS) emerging as the central platform for megasystems that will combine scheduling, billing, orders, monitoring and image management. The fully realized megasystem is itself still a ways off. However, certain critical components of it — legacy systems linked to new ones, patient data linked with images, all-in-one servers that store both DICOM and Web images, all layered with report and lab applications — are no longer the future. They’re here now.

The foundation
Integration is a big, fluid term that isn’t exclusive to enterprise systems. To examine those properly, though, you have to examine where they start. The building block is hybrid systems. Those are integrated, too; they actually comprise one of the first successful integration synergies and are also one of the highest-profile trends going. Driven by increased demand for more precise imaging and improved workflow efficiencies, dual-system synergies are well underway from most of the companies you’d expect and even some you don’t.

Consider the relatively new technology of PET/CT, which took only five years to seriously impact the market formerly owned by standalone PET. According to Frost and Sullivan, PET/CT sales represent 45 percent of the annual combined U.S. market of $481.2 million, and will likely overtake standalone PET sales within two years because of PET/CT’s value in oncology imaging.

Other promising combinations include another shaker in oncology, computer-aided detection (CAD). Such systems include CAD/mammo, CAD/CT and CAD/MRI. And now R2 Technology (Sunnyvale, Calif.), the first CAD vendor, has partnered with Sectra AB (Link?ping, Sweden) to produce the first CAD/PACS.

Susan Wood, Ph.D., V.P. of clinical and algorithmic development at R2, says, “Our system consists of an OmniCAD server that will house CAD applications for multiple modalities and disease states. OmniCAD sits on the PACS network, and then our viewing software will be integrated into the PACS workstation. We want to make CAD available to users any place, any time. We want to transfer the CAD information with the image data.”

The end result will be seamless access to R2 CAD applications for any Sectra PACS user. The FDA approval process, Wood says, “is well underway. The first applications that will be available will be for digital mammography [ImageChecker CT LN-1000] and for CT, for lung nodule detection.”

Sectra has been very busy lately, too. The company is also partnering with Viatronix (Stony Brook, N.Y.) to integrate its PACS with Viatronix’s V3D-Explorer (a 2D/3D MRI and CT workstation) and V3D-Colon (a CT colonography application). Omitting little with a slash, Sectra will also produce an integrated RIS/PACS with RIS Logic (Solon, Ohio).

On a true hybrid or other integrated system, the components are actually built together or otherwise share databases, rather than one merely being added on to the other. And that’s where the megasystem starts: with two or more systems conjoined to make a whole greater than the sum of its parts.

The RIS/PACS connection
The melding of information and image management gets the lion’s share of press these days in integration news, including (and perhaps most notably) radiology information systems (RIS) and PACS. There’s good reason. RIS/PACS is as transformational to image management as PET/CT and CAD are to oncology.

RIS/PACS is the first major component of the enterprise megasystem to have real legs. Quite a few vendors have jumped on the RIS/PACS train with truly integrated, brokerless hybrids. Novius/Sienet from Siemens Medical Solutions (Malvern, Pa.) and Centricity from GE Medical Systems (Waukesha, Wis.) were among the first. The latest include the single-database Entera from eMed Technologies (Burlington, Mass.), the Web-based PowerRIS/PACS from RADinfo Systems Inc. (Herndon, Va.), QDoc/Impax from Agfa HealthCare (Ridgefield Park, N.J.), and Medley from MTS-Delft USA (Aurora, Ohio).

Aside from its data tracking and consolidation capabilities, RIS/PACS is also pivotal because it prevents digital imaging exams from becoming “lost” in the archive due to human error. In 1999, The Institute of Medicine (Washington, D.C.) published a study claiming that up to 98,000 deaths result annually from preventable hospital errors; it is widely believed that megasystems of computerized records on shared databases with bidirectional communications will eliminate potentially fatal filing mistakes. RIS/PACS data integrated with hospital information systems (HIS) and computerized physician order entry would guarantee that all patient information remained solidly linked throughout the complete process, from initial presentation to diagnosis through treatment and results reporting.

Here’s how that might work. Hemant Goel, enterprise V.P. for radiology and clinical imaging at Cerner Corp. (Kansas City, Mo.), says, “A PACS workstation will never show you exams that are ordered or not completed. If for any reason an MRI exam did not end up in the PACS archive, the radiologist would never see that exam on his worklist and would not know it had been done.” In an integrated environment, he says, such instances are minimized by cross-checking applications between the RIS and PACS.

Another example, says Goel, “is that in a non-integrated RIS and PACS, someone could go into the PACS and modify an exam name or information,” rendering it incompatible with corresponding patient files on the RIS. In an integrated RIS/PACS, however, information that is changed on one side “will automatically change everywhere.”

Chris Wright, MCG Health System’s PACS administrator, has successfully integrated a 525-bed facility, children’s hospital, ambulatory care center and a variety of smaller users.

Everyone wants one
The megasystem of the future will have tons of useful information on it — useful to you, useful to creeps. Patient records are the motherload for thieves. They use them in identify theft schemes to commit fraud, or sell them to those who do.

The Federal Trade Commission says Social Security number theft is the fastest growing crime in America. FTC reports surged 88 percent in 2002, to 162,000 complaints. Seven people were charged in 2001 with stealing the Social Security numbers of more than 2,000 donors at a Chicago blood bank. They used them to obtain drivers licenses, establish bank accounts, and open credit card accounts with which they bought $2 million worth of plane tickets, hotel rooms, cars, computers and cell phones. In May of this year a similar fate befell 23 Pennsylvanians whose Social Security numbers were stolen from the Red Cross. Authorities in Detroit and Queens, N.Y., seized files full of hospital records from employees who stole and abused patient information. In 2000, an Internet hacker breached the database at the University of Washington Medical Center in Seattle and stole records containing 4,700 patients’ names, Social Security numbers, addresses, birth dates, height and weight.

One healthcare organization that will never become an FBI statistic is Mercy Health Partners (MHP), a seven-hospital system in Toledo, Ohio. MHP is in the process of networking all its facilities and referring doctors while eliminating film. So far, three facilities are online: St. Anne Mercy Hospital and St. Charles Mercy Hospital share a HIS, RIS and two archives with MHP’s main facility, St. Vincent Mercy Medical Center, where the data center is located. Some 350,000 imaging exams are stored there. All three facilities have RAIDs, workgroup servers, and DICOM servers. All their PACS and Web server equipment is from Kodak Health Imaging (Rochester, N.Y.), which assisted with the integration.

Leslie Beidleman, MHP’s PACS administrator, says that as part of the transition, MHP replaced its old medical record numbering system with a corporate identifier protocol. MHP’s Enterprise Access Directory (EAD) system from Siemens bypasses Social Security numbers altogether by filing patients under a corporate identifier number instead. Aside from the theft issue, Beidleman says, “a Social Security number is not reliable information, and you can’t always get that from people. You have the instance when a baby comes in who doesn’t have one, or someone from another country.”

– Leslie Beidleman, PACS Administrator at St. Vincent Mercy Medical Center, Toledo, Ohio

That’s not the only problem EAD solves. With three hospitals coming online at different times, Beidleman says, “there was a need to merge the information, and to make sure duplicate IDs weren’t being produced. If patients’ records are pulled up by their EAD corporate identifier and their images are taken under that number, then any facility can view all images, and they can be compared side by side.”

Name game
If there’s a downside to truly integrated systems, that would be the widespread bewilderment among potential customers about what integration really means, and how it can empower them. There’s confusion not only about terminology, and also about which products will work together and which ones won’t, and why.

For instance, many IT words have come to be used interchangeably even though they don’t mean the same thing. Much like the terms Internet and World Wide Web, integration and interfacing are at the top of that list.

“We differentiate quite a bit between them,” says Cerner’s Goel. “There are two levels of integration. If a radiologist can see the report and orders in the RIS and can look at the images all on the same workstation with a single log in, that’s visual integration.” The other type, he says, is “architectural integration — everything that happens behind the scene at the core system layer. That’s the kind of integration we talk about between RIS and PACS.”

Randall Swearingen, president of Swearingen Software, Inc. in Houston, offers this definition: If an application has exclusive rights to its own databases, it is most likely interfaced. If the application allows other applications to update its databases, it is most likely integrated.

All this fuss over semantics makes some customers gun-shy about choosing the components that best suit their needs. Swearingen says that buying a RIS and a PACS from different vendors isn’t all that risky, actually, if it’s done right.

“What we see typically happening is that facilities with really strong IS departments and technically competent radiology managers are much more apt to go best-of-breed, because they’re confident they’re making the right choices and they’re asking the right questions.” Facilities that aren’t technically literate or are less risk-tolerant, he says, “want to go with a ‘sure bet.’ But if you pick the best RIS and the best PACS, you can do homework that will lower your risk to zero. Some people are adverse to risk and just want something to work. And some people are competent enough to eliminate the risk themselves.”

Among those who aren’t are the ones who tell vendors no thanks, and head off to the local Best Buy to build their own applications with office software. Swearingen says, “It’s hard for them to see the complexity and the comprehensiveness of what a RIS does.” His company’s RIS, for example, can connect with most hospital and clinical information systems, PAC systems, voice recognition products, FAX machines, e-mail programs and pagers.

Mix’n’match
Combined systems are so much in demand that many vendors now sell multi-application packages, all of them termed integrated. Some really are, but some actually consist of products manufactured by different companies, and they’re not necessarily managed by software developed specifically for them.

Swearingen says many such systems “aren’t truly integrated. They’re interfaced just like any other product, and customers do have to make multiple phone calls” for support.

That doesn’t make them bad somehow, or unworthy. It just means it may take you a while to find the tech who can fix any problem that may arise with such a system. It could also mean your swap-out options may be limited, as some combined systems make it difficult to substitute components from other manufacturers.

Even when all the components originate from one source, sometimes the amount of programming actually written by the vendors for “their” multisystems is limited to things like user interfaces. In those instances, the nuts-and-bolts applications — in other words, the integration part — is outsourced.

A number of companies make very sophisticated gateways, interfaces and brokers for integrated systems sold by OEMs. One such company is Merge eFilm (Milwaukee, Wis.). It sells a product called MergeCOM-3 Advanced Integrator’s Tool Kit.

“It’s designed to help those who are building PACS, modalities, workstations and so on who need to receive or send medical images in DICOM format,” says William Stafford, V.P. of sales. “It’s for manufacturers, fundamentally.”

OEMs have their own programmers; everyone knows this. So why don’t they develop their own integration programs? Stafford says, “DICOM communication is not the place where the PACS company makes its money.” That would be interfaces — “the way a system does its storage hierarchy, or serves up images, or enables the routing of images. The actual dealing of the images themselves, while technically detailed, is not where an OEM’s high value add is. So they contract that out.”

Dell on speed dial?
Despite a market full of patchworked multisystems, Stafford is among those who say the do-it-all megasystem is the trend to watch.

“Right now we have specialty information systems for cardiology, nuclear medicine, radiology, and so on. Bringing those together and creating a more consistent environment within a hospital is the current thrust in the industry. What I’m hearing is that end-users are asking, ‘Why do I need a different system for each of the “ologies”? Why is there a different information system for each one?’ Effectively, it’s the same kind of data, with slightly different requirements in terms of presentation and workflow.”

Why, indeed, should any hospital have six separate information systems, says Stafford, when “somehow you’ve got to nail that stuff together to present a coherent picture about one patient?” An excellent question.

Here’s the answer: A few little things need sorting out before the integrated megasystem can become an industrial reality. One is the standardization of components and the elimination of proprietary interfaces. Another significant issue, Stafford says, “is the age and stage of the installed base of equipment and applications. Most of the modalities sold today do DICOM store and DICOM print, and many even do worklist management. A relatively small proportion do the modality perform procedure step, which the modality does after a procedure to tell the PACS or RIS what was performed.” But the main roadblock, says Stafford, is that “not every piece of the institution that has to be woven together is at the same level. They’re at different ages and stages of development, and frequently they’re updated or upgraded. External components and third-party pieces such as black boxes are necessary to effect integration. That’s true in almost any IS environment today.”

And so for now, the truly integrated megasystem remains around the corner. But just barely.

Some institutions have forged ahead anyway in the quest for the Big One. What does integration look like when all the pieces come together? A good example is MCG Health System in Augusta, Ga.

Chris Wright, MCG’s PACS administrator, says, “We are the academic medical center for the state of Georgia, so our referrals reach pretty far.” The main facility, the 525-bed MCG Healthcare Inc., serves the whole state and parts of South Carolina. Its radiology department is 65 percent digital, but all components of the health network are using its Siemens PACS. The intranet is an Ethernet network, and the archive uses a tape library from StorageTek (Louisville, Colo.).

“We started with the original archive server in November 1998, with ER and computed radiography,” says Wright. “Our Children’s Medical Center opened six weeks later, and it was preplanned to coincide. Since then we have archived all CTs, MRs, and ultrasounds.” MCG’s first archive system was an ultrasound mini-PACS acquired in 1996. Its two 150-MOD jukeboxes filled up within the first eight months, and the 300 disks required constant juggling. MCG realized they needed something beefier and more automated, and subsequently migrated everything to a tape archive. It currently holds 13.7 million images.

In addition to DICOM image management, MCG also has a Web server. “Recently we started allowing extra-campus access,” says Wright. “We have partnerships with rural hospitals where we’re helping out with specialized readings. They send images to us through a secure socket-layer connection. We also do simple telerad for on-call at night. The radiologists have access to the Web server from home, and they can help the residents without having to get in the car and come back down here.” MCG’s Ambulatory Care Center consists of 80 outpatient clinics that will soon be online as well.

“We have a PACS broker that converts HL7 messages from our HIS and RIS to DICOM messages. This allows for worklist management applications to provide appropriate patient demographics to a modality for a specific exam. This device makes it possible for the technologist to register the patient and place an electronic order for the exam to be performed in the RIS, and then simply choose the patient’s exam from a worklist on the modality prior to imaging. A single registration event can supply patient demographics to multiple imaging devices, eliminating typographic errors in patient demographic data entry. The broker also provides verified results (imaging interpretations) to workstations and associates the results with the correct imaging study.

“We store an average of 16,000 images a day,” says Wright. “We support our emergency room completely with digital, and all of our ICUs. About 90 percent of the rest of the in-patient portable exams are done with CR. Children’s Medical Center is entirely digital with the exception of the operating rooms, because we haven’t found a solution for displaying images that everyone’s happy with there.” Well, nobody’s integration is perfect. But MCG’s is pretty darned close.