Feature Story

Pain, then gain: one laboratory’s automation story

December 2000
Karen Titus

Inside the small lab conference room at the University of Chicago Hospitals and Health System, all is calm as Ronald W. McLawhon, MD, PhD, quietly describes the automation setup in his laboratories.

Just outside the door, where lab personnel are busy processing a portion of the 4,000 patient samples that pass through the laboratories each day, all is calm as well. Across the hallway, in another section of the labs-calm. Ditto the atmosphere in the adjacent offices.

Think of it as the calm after the storm. Bedlam reigned here just a few short years ago, when UCHHS embarked on an ambitious-some might say "brave"-plan to automate its laboratories, boost its outreach business, and save a few bucks in the process.

Today, it’s apparent UCHHS laboratories did all that and more. Much of their success can be credited to the modular automation system now in place, which handles routine and stat chemistries, therapeutic drugs, drugs of abuse, and proteins. The modular approach is often touted as a fitting alternative to total lab automation, which indeed was the case at UCHHS, notes Dr. McLawhon, director of Regional Laboratory Services, medical director of Hospitals Laboratories, and associate professor in the Department of Pathology.

It would be nice, then, to report that automation enjoyed a flawless debut at UCHHS. Not surprisingly, it didn’t.

As with any major implementation project, the move to modular was arduous. Smooth sailing is the stuff of storybooks (and sales pitches); in the laboratory, such an enormous shift quickly becomes a tale filled with hassles and heartaches, resentment as well as rejoicing, with a few unexpected twists thrown in for good measure. In an interview with CAP TODAY, Dr. McLawhon and laboratory manager Steven Zibrat, MS, MT(ASCP), recalled the circuitous road that led to lab automation and offered advice to other labs heading in the same direction.

Like most laboratories, the clinical labs at UCHHS have been buffeted by the usual machinations of contemporary medicine: alliances formed and dissolved; takeovers tried and failed, then tried again; the incursion of managed care; integrated delivery system face-lifts; and antipodal pleas to make more money with fewer staff.

In response, the labs looked to their outreach program, UC MedLabs, as a solution to at least some of these ills. The program had been in place since 1986, but it had waxed and waned over the years. In 1995, "We made the decision to rejuvenate UC MedLabs and step up our outreach efforts," says Dr. McLawhon, which they hoped would offset the 20 percent drop in inpatient test volume at UC Hospitals between 1993 and 1995. They also decided to reposition themselves as the primary lab service provider for the newly formed and rapidly evolving UCHealth System. "Prior to that, there were a number of different lab providers for the same physician offices and various hospitals," Dr. McLawhon says. "We decided we had to go where the work was going."

By 1997, the Hospitals Laboratories, which served three hospitals at the campus medical center, and MedLabs were brought under the same umbrella, organized as Regional Laboratory Services. The entity provides all diagnostic and pathology services for all UCHHS facilities, affiliated and nonaffiliated hospitals, and area clinics and physician practices.

The simultaneous buildup in lab outreach testing and drop in inpatient volume placed Dr. McLawhon and his colleagues in the middle of an economic conundrum: how to downsize while building the business. "That’s when we started thinking about automated solutions," he recalls. At the same time, the labs began evaluating other changes they would need to make to attract and accommodate outreach business, including adding a central customer service center and transforming the clerical order entry shop into a technical work unit.

UCHHS’ move toward automation appears to have, in retrospect, a rather lurching quality to it. "We got halfway through a couple steps and changed directions a couple of times," Dr. McLawhon concedes. They first looked at total lab automation, then in its early stages of development. After lab faculty and management staff and hospital vice presidents made several on-site visits to view different types of TLA setups, UCHHS began inching toward TLA.

"Certainly we had a mixed camp," Dr. McLawhon says. "Some of the people thought [TLA] might be a possibility, and then there were those who said, ’You don’t need it, it’s going to erode our academic disciplines, you’re going to be throwing various clinical subspecialties all together on one line.’

"The concerns were understandable, but they had to be weighed in the context of economic pressures and the goals of the entire enterprise-not just those of the labs," he says. "So we had all those barriers to deal with internally."

Then there were the physical barriers. Though the building that houses the labs was erected in 1983, it’s surprisingly antediluvian, an example of what Dr. McLawhon calls "the old school of laboratory design, with lots and lots of walls and lots and lots of hallways"-including a couple of prominent firewalls that ran right down the middle of the labs.

Finances, surprisingly, were the least problematic. The capital budget committee approved the project, and the dollars were just waiting to be spent to renovate space to accommodate the newly available Roche/Hitachi Clinical Laboratory Automation System, or CLAS.

And then the dollars went away. With the CLASfootprint so large and the required renovations so extensive, UCHHS ran into other barriers-this time bureaucratic. Because of the lab’s proximity to patient care areas, UCHHS spent months navigating municipal and federal codes-some arguably archaic. "We had to wait for building permits to be issued to move forward. And the budget committee wouldn’t release the capital dollars to install CLAS until this was resolved," Dr. McLawhon explains. Eventually the labs prevailed, but by that time the capital cycle had passed for another year and the funding had been allocated elsewhere.

With the dollars gone, "We just decided, OK, we’re not going to pursue total lab automation at this point. That’s when we got introduced to modular," Dr. McLawhon says. "We figured it would take less space, cost less, probably would accomplish a lot of the same things, and we didn’t have another year to wait. And if we did want to go to the point of total lab automation, we could plug the modular system into that," since the modular system they were considering was slated to be the heart of Roche/ Hitachi’s second-generation TLA, CLAS-II.

The labs agreed to serve as an evaluation site for the Roche/Hitachi MODULAR Analytics and Preanalytics system in August 1997, to be the first to take on the new combined technology in a clinical setting. Dr. McLawhon and Zibrat knew the peril of so ambitious a project. "Being at the leading edge means you’ll often find yourself at the bleeding edge," Dr. McLawhon admits. Installation began the following summer.

In between, Dr. McLawhon and his colleagues kept busy with a series of mind-boggling changes in the labs. For starters, three aging analyzers were replaced. "They were getting a little long in the tooth, and we needed to make an interim switch before the MODULAR system came online. So we effectively changed instrumentation twice within the span of a year," Dr. McLawhon says.

"We also did a renovation project to accommodate the integrated Analytics-Preanalytics line that literally had this place in shambles for the span of nearly five months," he continues. "And we had to finish it on an accelerated timetable, because we had CAP and JCAHO doing their inspections at the same time." Lab personnel found themselves sharing space with construction crews jackhammering through the floors, knocking down walls, and creating new doorways. Recalls Zibrat: "You’d come into work one day and your analyzer would be over here. You’d come into work the next day and it would be on the other side of the room. Fortunately, the staff was just phenomenal in dealing with that," maintaining a 24/7 testing schedule and performing the initial evaluation on the Analytics portion of the MODULAR system.

The labs also participated in an international multicenter trial to evaluate the systems. Prior to installing the final clinical configuration, the labs assessed a prototype as well as an evaluation unit as Roche and Hitachi modified the system’s hardware and software. The clinical release units of the system, both the Analytics and Preanalytics, were brought into the lab in April 1999. The labs also had to install a new interface to link their customized Sunquest LIS to the new system and to yet another new element, the Data Innovations Instrument Manager workstation. This complicated job was made even more so by Y2K preparations at UCHHS and Sunquest.

Finally, in September 1999 the lab placed MODULAR Analytics into routine clinical service. And by last December, after completing the evaluation of the Preanalytics module, the labs went live with the integrated system.

"And then we crashed and burned," says Dr. McLawhon with a laugh.

"You go through an evaluation, and it’s a very sterile environment," he explains. "Everything is well defined, and you don’t challenge it necessarily with real-world specimens." Although the UCHHS labs ran practicability studies during the evaluation, using actual specimens and orders, "When you put it into 24/7 production, with everyone working on it and not just the principal evaluators, you’re throwing in some monkey wrenches not there before. Things are bound to fall apart." The problems were "a combination of everything," he says-a new system, new software, and new working procedures, typical of any major shift in routine.

Seemingly small matters, such as specimen tube size, type, and labeling, proved problematic at first. The UCHHS labs wanted to switch from glass tubes to plastic ones. "We wanted to prevent possible breakage from glass that could occur inside the modules," Zibrat explains. "Plus, plastic is uniform, as opposed to glass, which can vary as much as a quarter of an inch in height, for example"-no small consideration when a system is automated. "When the grabber comes down to get the tube, it has the potential to rip the top of the tube right off if it’s not at the right height," Zibrat says. "Or it might not be able to recap the tube properly, either."

The labs began shifting their outreach clients to plastic tubes several months before the system went online, but found their in-house customers to be less accommodating. "There’s always someone in the hospital’s clinics who still has 15-mL glass tubes squirreled away from 20 years ago," Dr. McLawhon says.

Problems with labeling also cropped up in the course of daily routines. "When you’re evaluating the system, you’re very careful to make sure every label is perfect," says Dr. McLawhon. Such exactness is needed for an automated system, which cannot read bar-code labels facing in the wrong direction or that are crinkled or otherwise improperly placed or damaged. "If you’re not precise in labeling practices at the bedside, or if you receive specimens with multiple layers of labels wrapped around them, the tubes will not spin properly or be scanned by the system, and an alarm will sound."

Which in itself was problematic. While the manual procedures weren’t seamless, he notes, "At least you have that down. Everyone knows the process. But when you make the switch to an automated system, people don’t know what to do at first. If an alarm goes off, everyone just stands around and stares, and they don’t know what to do or they’re scared they might do something wrong."

Making the jump to automation is more complicated than simply changing instrumentation, Dr. McLawhon contends. "I’ve gone through instrument changes, and they can be wrenching," he acknowledges. "But this changes your entire process and approach, your workflow and mindset. And you constantly have to be on top of things. If something goes wrong in the process, you’ve got to take the time to figure out what’s going on. And that’s not easy for everyone to do, especially since it’s a whole new way of solving problems."

Adds Zibrat: "It’s not just like walking up to the system and saying, ’Well, it’s just a big analyzer.’ You have to hit everything at the right time, and if somebody misses their cue, everything gets backed up."

That said, when the system is fully operational, "You really can push things through in a rapid fashion," Dr. McLawhon says. The system now processes up to 450 specimens per hour and generates 5,800 test results per hour.

It’s not that Dr. McLawhon and his colleagues didn’t anticipate problems arising with the shift to automation. "But you don’t get a true sense of the magnitude of those problems until you start to work through them," he says.

In particular, Dr. McLawhon says, both UCHHS and the vendors may have miscalculated on the training aspect.

Key operators at the UCHHS labs were trained on the Analytics and Preanalytics portions separately, for example, but were not trained to handle a combined and fully integrated automation workcell. Likewise, the vendor "had an expert on the Analytics system and another for Preanalytics, but not someone who could readily bridge the two," says Dr. McLawhon. Ultimately, he says, the vendors used the experiences of his lab to revise their approach. "We all learned from this process." At the same time, other challenges remain. The system’s autoverification system remains underutilized, Zibrat notes, because the processor capacity of the aging Sunquest LIS cannot keep pace with the rate of information transfer from MODULAR. The Sunquest hardware is set to be replaced soon.

The system’s stat interrupt feature has also been largely idle. "The technologists like to keep a hands-on approach and always see the sample. So they don’t use the feature like they should," he observes. "That’s just one more example of the mindset that needs to be broken when you move to automation."

"As well as the system works, and as flexible as it is, there are some real-world problems you encounter when you make the switch," adds Dr. McLawhon.

Touring the labs now, it’s hard to imagine any of those difficulties ever existed. The construction crews and dust have vanished, no alarms sound, and laboratory personnel tend the sleek system with quiet efficiency. "We’re not shy about what we throw at this system," says Dr. McLawhon. The labs perform 2.7 million billable tests each year, almost 50 percent more than what they did in 1997. They serve approximately 110 outreach locations, a category that has also seen dramatic growth in the last two years. Revenues from off-campus locations increased tenfold from reference lab outreach activities and laboratory management services contracts.

None of these increases could have been managed without automation, Zibrat says, given that "We had the constraints of trying to control costs at the same time, and had to grow the business while keeping the staff at status quo." Further automation was slated for the postanalytic end, with the lab planning to add a decapper/sorter/archiver module, the Roche PSD1 task-targeted automation system. In addition, two E170 immunoassay modules will be incorporated into the Analytics system when they’re available in mid to late 2001.

Dr. McLawhon has at least one other goal of his own. As he and Zibrat ponder future expansion in the labs and the growth of automation, they eye two solid-looking walls near the system.

"We do have those walls to contend with," Zibrat says.

"At least for now," Dr. McLawhon replies. "My fantasy is still to bring those walls down."

Karen Titus is CAP TODAY contributing editor and co-managing editor.