Feature Story

Turning up the volume on lab automation

June 2002
Eric Skjei

After a period of slow growth, the lab automation market is again showing signs of life.

Interest in automation has blossomed, thanks to the chronic shortage of medical technologists, increasingly attractive returns on investments, ready availability of proven automation solutions, and evidence from several installations in large and small labs that automation can improve not only the bottom line but also quality, patient safety, and workflow.

Wary of committing to a costly total lab automation overhaul, many labs are choosing to focus on the front end of the test cycle—sorting, aliquoting, and centrifuging.

"In our customer base, the bulk of the automation installations have focused on front-end automation versus total lab automation," says Lee Green, CEO of Triple G Corp. "These are very business-oriented labs, and they have found that, after crunching the numbers, this is an approach that makes sense."

Early adventures in total lab automation in the mid- and late-1990s often took excessively long in achieving returns on investments. Today, labs that are focusing on the front-end process are seeing "costs and payback times that are coming down into the two-year range, which is much more acceptable," Green says. ROIs of this scale make lab automation attractive for not only mid-size labs but also, as the ROI time frame continues to shrink, for smaller labs. In some cases, front-end automation is being extended to include an instrument or two in what might be defined as a modular strategy, one that allows labs to add automation stations and functions as budget and other resource constraints permit.

As a result, the market for automation is picking up. Lab automation pioneer Rodney Markin, MD, PhD, says the company he founded, Lab-InterLink, will sign 15 contracts this year. "There are probably 50 to 75 automation sites out there live right now, and depending on the size of the system, there is a potential total market of perhaps as many as 3,000 clients," says Dr. Markin, professor and vice chairman of pathology and microbiology and associate dean for clinical affairs at the University of Nebraska Medical Center, Omaha.

Misys Healthcare Systems (formerly Sunquest Information Systems) also sees a growing appetite for automation interfaces among its laboratory information systems clients. From 1996 to 1998, the company handled one automation interface. In 1999, that number grew to four, in 2000 to five, and in 2001 it doubled, to 10, reports Elinore Craig, Misys’ director of laboratory and financials product management.

"We sense that the automation market is on the verge of exploding," she says. In her view, the driving forces are largely economic and pertain to the cost advantages of being able to replace full-time equivalent employees with robots. "If only because med techs are in such short supply, labs must find ways to dedicate their skilled workers to skilled tasks, and if a robot can handle the less-skilled tasks instead, then they need to look at this opportunity," Craig says.

Despite the welcome signs of growth, challenges remain. Even when an automation system is implemented in phased steps over time, the financial investment is significant. Though installation, training, and maintenance are not as burdensome as they once were, they still can be costly and disruptive. Finally, though interfaces between automation systems and the LIS generally perform well, the software systems involved can and should operate more harmoniously.

In the last few years, the role of the LIS vis-à-vis the automation system it interfaces to has been transformed.

"When we first started doing automation interfaces five or six years ago," says Misys’ Craig, "we thought that perhaps the automation system would want all its direction and information from the LIS, that the automation system would just be a dumb robot, basically a conveyor belt, and that the LIS would have to run that belt."

Many automation systems, however, are so intelligent they need only minimal information from the LIS and effectively control all the processes and data once the tube is delivered to the lab. In short, the automation system, not the LIS, is calling most of the shots within the test process in an automated lab, including, for example, managing rules activation, reflex testing, and delta checking. In many cases, the role of the LIS is simply to uniquely identify each tube, communicate the test codes associated with that tube, and acquire and file the results at the end of the process.

As the two systems interact, unless they are carefully coordinated, questions can arise about which system should handle what chore. And even as these questions are answered, the two systems can create redundant data requirements, parallel and even overlapping databases, and inefficiencies that can compromise the performance of both systems and the laboratory.

Call for an LOS

A few of the more forward-thinking leaders in the lab automation industry are inherently uneasy about the coexistence of automation system and LIS. The two should be more tightly integrated, they say, as has been done in other industries.

"If you look at complicated business systems in manufacturing as an example, you will see that the manufacturing information systems include process control along with the information about the product," says Dr. Markin. "If you go to GM, for example, and look at how it manufactures trucks, you will find not only the process information and all of the history of how a given vehicle was built [residing] in the system managing that process, but also information about who it was built for, who owns it, when it was ordered, was it a special order, a routine order, and so on."

The parallel with the laboratory is self-evident. The potential problems that can arise in the absence of an integration strategy are perhaps less obvious.

"The issue is that in the laboratory, we now have two parallel pathways of information technology that I believe should be fused into one," says Dr. Markin. On the one hand, the LIS handles orders, collates and reports results, assists with quality assurance, and in some areas handles billing. The parallel system—the lab automation system—represents a "huge overlap" with the LIS in terms of the elements of its database, but it also contains a "distinctly different, process-oriented database with information about the history of the specimen, including its physical location, historical tracking, its physical state—whole blood, plasma, frozen, not frozen—and so forth," he adds.

If, says Dr. Markin, you were a process-control expert from GM and you walked into an automated laboratory and saw these two largely overlapping systems working in parallel, you would wonder if the process could be improved. And in Dr. Markin’s view, you would be right.

"I think that the natural evolution of lab automation is to take the [laboratory] information system and bolt that together with the process or the physical automation system," he says. If you do this, you get what he refers to as a laboratory operating system, or LOS. Like any other traditional computer operating system, an LOS would run the hardware and software components of the test process and provide a platform for more advanced applications and databases.

The barrier to this level of integration of LIS and automation system, say those closest to both worlds, is not so much technical as cultural. The level of integration it presupposes would require greater collaboration between the LIS and in vitro diagnostics industries, and if history is any measure, that is not likely to happen anytime soon.

"I always thought there was a terrific opportunity for these two domains to be more integrated than interfaced, but it’s never happened," says Bruce Friedman, MD, professor of pathology at the University of Michigan Medical School, Ann Arbor. Why not? "For one reason, the IVD industry has traditionally been very large compared to the LIS industry, which has tended to be a world of smaller shops," says Dr. Friedman. "As a result, I think the [IVD manufacturers] just don’t feel compelled to cooperate closely with the LIS vendors."

There are also, suggests Dr. Friedman, the underlying cultural differences. "The IVD space is mainly controlled by engineers, who work under very heavy scrutiny from the FDA," he says, "and who tend toward a very fastidious engineering mentality that prohibits bringing products to market until they are fully tested." By contrast, the LIS world long ago accepted as fact, as have most software developers, or engineers, that software never goes to market in a fully perfect form, and that theirs is a world in which products are always evolving and being revised and upgraded, even while they are being delivered to and used by customers.

Integration at the level Drs. Markin and Friedman envision raises other questions, including, for example, that of performance. Says Sue Masarik, laboratory manager at Chicago’s Northwestern Memorial Hospital, "There’s a lot of overhead to big integrated, number-intensive systems, and when you hook something that is as computer-intensive as an LIS can be into it, you very often see performance hits and slower response speeds, even with the fast computers and networks we generally have available today."

Beating the ROI game

In the meantime, a laboratory that wants to deploy automation must make its case on interfaces and projections for returns on investments. Most laboratory professionals realize that meeting or beating those estimates can make or break their management and their future. The good news is that there are a growing number of clear-cut cases where laboratories have achieved compelling investment returns in a complex, staged lab automation installation.

Take, for example, the laboratory managed by Leo Serrano for the West Tennessee Healthcare system. West Tennessee Healthcare, based in Jackson, Tenn., is an integrated rural organization located midway between Nashville and Memphis. With a service population of more than 500,000 and an aggressive outreach program with about 1,000 clients, the lab handles sizable test volumes.

When Serrano arrived in Jackson six-and-a-half years ago, he was handed the job of doing more with less—specifically, building an outreach program with FTE constraints. "I was hired with the understanding that we would expand our outreach and increase our market share," he says. Serrano, executive director of laboratory services, realized immediately that the lab, then organized in a way that was more suitable for the late 1960s, needed a significant overhaul if it was going to be able to handle such growth.

Relying on his background in management engineering, Serrano conducted a workflow analysis and came to an initial conclusion that hematology was one of the major bottlenecks. After reviewing several possible automation solutions, he invited Sysmex to study the lab and conduct a workflow analysis. Sysmex’s conclusions closely matched his own, lending credence to the vendor’s proposal to incorporate the laboratory’s analyzers into an automation system.

"We took the proposal to the board, promising them that it would allow us to reallocate three FTEs, which is a very real benefit in a rural location like ours where it is particularly difficult to recruit med techs and other lab staff," Serrano says.

The success of the hematology line has far exceeded expectations. Instead of saving only three FTEs, the lab was able to reallocate five FTEs and generate a positive ROI in 23 months. The lab also improved turnaround time for routine testing and absorbed increased test volumes. "At the time we were doing about 400 CBCs a day," recalls Serrano. "Today, we’re doing between 900 and 1,000 a day."

With this initial success under their belt, Serrano and his team began to consider other areas in the lab where automation might pay off. "We then took a look at the ’other side of the room,’ as we call it—chemistry, immunochemistry, and coag," he says. West Tennessee has one of the premier cardiac programs in the state so it performs a lot of anticoagulant work—about 500 samples a day, according to Serrano.

Once again, his group conducted a thorough workflow analysis and reviewed several vendor offerings. They decided to implement a six-station system from Lab-InterLink that promised to consolidate much of the immunochemistry work, which was scattered throughout the laboratory.

The lab knew what it needed in the new automation system. "We wanted a system that was totally open, that did full front-end processing, including not only centrifugation but decapping," Serrano says. "We wanted a system that was smart, that could communicate well with our LIS, and that was rules-based."

The product had to be able to base rules on the automation system or the LIS, or both. Lab-InterLink software includes rules, as does the lab’s Meditech LIS. Before the project began, Serrano wondered how well the LIS and the automation software would mesh.

"If you’ve had any experience with automation, that’s the big scare—how the LIS and the process-control system will work together," he says. In his case, the worry was unjustified. "The LIS portion of the installation turned out to be the easiest," he says. "Between Lab-InterLink and Meditech, we were able to create a combined system that thinks pretty much like a tech."

The system permits autoverification and release of results, knows where the sample should go after it reads the bar code and queries the LIS for additional necessary information, knows where the specimen must be decapped, understands a user-defined centrifugation time, and much more.

Cost was, of course, a concern, especially since the state of Tennessee requires a Certificate of Need for projects that exceed $1 million. To avoid the CON requirement, the lab split the project into two phases, each costing less than $1 million. Phase one, involving front-end processes and four stations plus back-end archiving, was completed in December 2000. Phase two, now underway, will allow immunochemistry and coagulation specimens to be sampled directly from the automation track and should be completed by the end of the year.

Serrano and his colleagues again beat their ROI projections. At the project’s outset, the ROI was estimated at 31 months. "We actually achieved that sooner than we expected and are now at 27 months," Serrano says. "Overall, automation to date has saved us about 11 FTEs, which we have reallocated."

Turnaround times have improved as well. From the time a basic metabolic panel specimen reaches the lab to when results are available in the emergency room, for example, is now 19 minutes, and a CBC turnaround time with differential is 16 minutes, and without differential nine minutes. Serrano makes no secret of the fact that he is pleased with the results of his automation endeavors, his internal and external partners, and, most of all, his staff.

The process is the point

Being able to replace costly, imperfect humans with reliable, accurate, depreciable machines—and in so doing improve patient care—is, of course, the main appeal of lab automation technology. And while it’s true that most labs undertake automation projects primarily to reduce and reallocate FTEs, some argue there are other benefits that are not as self-evident but are equally, or perhaps more, compelling.

At Northwestern Memorial Hospital, a laboratory team under the direction of Kenneth Clarke, MHA, MBA, is pursuing a vision of automation that is more broadly focused than just ROI. "We automated for one basic reason, one that had a lot of unknown and perhaps unknowable benefits, and that was to improve our process," says Clarke, who is administrative director of pathology laboratories.

In Clarke’s view, many of the most powerful benefits of automation cannot be known in advance. "Any article on automation spells out the same four or five benefits—FTE efficiency, improved consistency, better quality, faster turnaround time—and we had all those things in mind too," he says. "But we think there are a whole set of things that we can’t know in advance, benefits that will help us, the hospital, patients, and all the users of the lab, especially physicians."

Clarke rejects the strategy of piecemeal, phased, staged, or modular automation.

"We disagree with people who think front-end or modular automation is the way to go," he says. "We understand the benefit, but we don’t think you can really improve the process by just dealing with the parts of it." These strategies still require some sort of manual intervention or system-to-system handoff, he adds. "We’ve toured labs that have spent millions on automating some components but not the $50,000 needed to tie those components together."

Today, one year into its project, the Northwestern Memorial Hospital laboratory has one Beckman Coulter chemistry line that is up and running, and a second, two-track system with an operational coagulation line and a hematology line that is six months away from going live.

So far, one of the unpredictable benefits pertains to the back end of the process. Refrigeration installed as part of the automation venture has made it possible to automate the add-on testing that typically takes place after an initial test series has been ordered—and, here, the automated process does not need human intervention.

Says Mark Wasielewski, specimen-receiving coordinator: "When we put an add-on into the system, it automatically goes to the refrigerator, pulls out the specimen, runs the test, and autoreleases the results. No one at all needs to know about it, other than the individual who actually ordered the test."

After Northwestern implements an order-entry system hospital-wide next year, it will be possible to activate this function remotely, from the floors or even off-site. The system is already being used in Northwestern’s emergency department. "As we speak," Clarke says, "someone in the ED could be ordering a test and the system could be finding that tube in the refrigerator, pulling it out, sending it down for testing, running the test, releasing results automatically back to the system, then sending the sample back to the refrigerator, all without any of us in the lab even knowing that any of this is taking place."

Inspired by the success of this approach, the laboratory has taken it a step further. It is developing a "draw-and-hold" process for the emergency department. "As a patient comes in, the emergency department will draw blood, and we will put that specimen on the line before any tests are ordered," Clarke says. "Then, when a test is ordered for that patient, this entirely automated add-on process will be called up—the system will pull the sample, already centrifuged, out of the refrigerator and quickly run it, and that will cut turnaround time significantly." In fact, the laboratory has already cut its turnaround time to the emergency department for troponin from 90 minutes to 45. With this draw-and-hold process, Clarke and his staff foresee being able to cut that time in half.

"This has really eased some of our workload," says laboratory manager Klaus Nether, "because we don’t see anything in terms of the add-ons—we just pull our worksheets and the test is already on it and the specimen is already here."

Along with the unpredictable benefits come more predictable ones, and Clarke’s team points to two they have already seen. The first is a reduction in errors.

"People make mistakes," says Wasielewski. "As we move forward and bring more automation into the hospital, especially with respect to the labeling that will become available in the order-entry process, we will see fewer people handling the specimen." Tubes will arrive in the laboratory already labeled, and staff will simply put them on the line. By contrast, nurses now fill out a lab requisition and hand label the tube, a step that can introduce error. Then they send the specimen to the lab, where a staff person must key in basic information and relabel the tube with a bar-code label, another step that can introduce errors. "The fewer people who have to handle the specimen," Wasielewski says, "the fewer the number of errors that will occur."

At the same time, being able to enter orders on the floors will help shift workload out of the lab and into other parts of the hospital—making it possible for the lab to take on more business. "The automation line and the instrumentation we have now are capable of processing a lot more throughput than we’re pumping through it today," Wasielewski says. "Currently we process about 2,000 tubes a day, but we’re capable of five times that volume."

Lab manager Nether also names increased throughput as a benefit of the lab’s automation investments. "We will be able to bring more testing into the lab," he says. "And even then, I don’t foresee that we will need to hire another FTE since a lot of the workload from the front end has been taken care of."

Rough spots remain, however. And some are caused by functions not yet being available in the LIS, which is scheduled to be upgraded soon. "We have to validate that every tube that has been ordered has arrived," says Vida Jesewitz, LIScoordinator. But at the moment, "certain pieces of information we need to do that—receive times, collect times, and so on—are not present in the LIS when the tube arrives. Staff members still have to go into an order-entry/modification function on our system and manually enter what we have and have not received."

Eric Skjei is a writer in Stinson Beach, Calif.