College of American Pathologists
Printable Version

  Feature Story


cap today

To lower costs, one lab takes automation to task

July 2003
Karen Southwick

Like many hospital laboratories, the one at Abington Memorial
Hospital in eastern Pennsylvania was caught in a crunch: demands from management to lower its costs and competition from a reference lab down the street.

In 2000, the Abington laboratory participated in a benchmarking study using HBSI Action and found, to its dismay, that “we weren’t meeting our targets [compared with peers] for hours worked per billable test and costs per billable test,” recalls Kathryn Durr, administrative director of pathology. Moreover, Abington was adding outside clients, such as nursing homes, to boost volume and compete with the reference lab, but its existing instruments couldn’t keep up with the demand.

“We needed to look at better instruments and processes to support taking on additional clients,” she says.

Abington’s pathology department chair, Herbert Auerbach, DO, speaking in May at the Dark Report’s Executive War College, described the lab as having “run into a wall.”

“We couldn’t increase our billables per worked hour. We were at the 50th percentile using HBSI data, and we weren’t making any improvements,” he told his audience in New Orleans.

At the same time, hospital administrators at Abington Memorial, a 508-bed comprehensive medical center, were pushing the laboratory to cut costs. The lab runs more than 3 million tests a year, mostly chemistry, of which 42 percent are inpatient and 58 percent are outpatient.

“We thought we were pretty efficient in running assays,” says Dr. Auerbach, who also spoke recently with CAP TODAY. He says the HBSI data were skewed because Abington has a higher procurement cost for outpatient tests than most of its peers because it provides extensive phlebotomy services to its clients. But the hospital brass disagreed. “They were lumping all our costs together and looking at us versus our peers, and they just wanted to get [those costs] down,” he says.
Meanwhile, it was the same story everyone in the lab community faces: Payers were ratcheting down fees. “Nobody wants to pay for lab tests,” Dr. Auerbach says. “A lot of our clinicians and patients regard it as a commodity like water. They expect it to be there when they turn on the faucet, and everybody wants to pay less.”

The pressures to cut costs while building volume led to a multiyear, two-pronged effort in which Abington Memorial’s laboratory reorganized its chemistry, hematology, coagulation, and urinalysis sections into a single core lab and followed that up by automating the majority of its tests there. With lab automation, Dr. Auerbach says, “we hoped to do it cheaper, faster, and better.”

Task-targeted automation
In planning for the automation, which was executed in 2001 and 2002, Dr. Auerbach and Durr visited other automated labs and took with them some of their lab personnel to help with the evaluation. “We drew up a list of the vendors we thought had offerings, invited each to come to our lab, do a workflow analysis, and make a proposal as to how their equipment would fit,” says Dr. Auerbach.

Abington decided to bring in task-targeted automation rather than total lab automation. The idea was to automate one section of the laboratory at a time. Each section would get its own automated instruments targeted to its own workflow. That kept the price tag lower than it would have been for total lab automation and gave Abington the flexibility to negotiate with several vendors rather than one.

“Task-targeted automation offers much better use of technologist labor,” Dr. Auerbach says. “We didn’t want to automate a bad process.” The aim was to simplify processes with automation to reduce errors and turnaround time and expand the test menu.

The core laboratory represents 88 percent of lab charges, Durr told CAP TODAY. “We targeted areas for automation where we could get the most bang for the buck,” she says. “In terms of sheer volume,” Dr. Auerbach adds, “when you do chemistry and immunoassay, you’re doing 70 percent of your lab. Add hematology and you’re at 80 to 90 percent.”

Hematology was the first to be automated. Abington purchased a Cell-Dyn WorkCell system from Abbott Diagnostics; it consisted of two Cell-Dyn 4000 automated cell counters, the slidemaker/stainer module or SMS, a recirculating track that transports the specimens using a rack-based conveyor, and the Navigator data manager. The incremental cost for the automation—excluding analyzer equipment Abington would have purchased anyway (for example, the Cell-Dyn 4000)—was about $200,000, Dr. Auerbach estimates.

In evaluating expenditures for automation, Abington estimated that for every $100,000 spent on automation it would have to save about $35,000 in costs annually to compensate for the capital expenditure amortized over five years and the estimated cost of a service contract. “To translate that into how many worked hours we would have to save, we used a simple formula,” Dr. Auerbach says. There are 8,760 hours in a normal year. A regular work year is 1,960 hours. “Therefore, for example, if a piece of automated equipment could save one position across all shifts seven days a week, that would result in a savings of 8,760 divided by 1,960 or 4.5 full-time equivalents,”
he explains.

“In the core lab area of chemistry and immunoassay we sought an integrated solution,” he says. “We wanted an analyzer or system of analyzers that would consolidate much of the testing onto one platform.” Abington selected the Bayer Advia WorkCell, which included two Advia 1650 chemistry analyzers and two Centaur immunoassay analyzers. “Bayer was the first to offer an automated integrated chemistry and immunoassay system. Before this, laboratories that attempted to integrate chemistry and immunoassay testing had to do it with customized automation,” he says.
This trend of integration is becoming more common. The analyzers are linked by an oval track that routes the specimens to the proper analyzer based on the ordered tests. When the tests are completed, the sample manager can hold the tubes for reflex testing or sort the tubes for archival or for off-line testing. “By combining two random-access high-volume immunodiagnostics analyzers with two high-volume chemistry analyzers, we were able to streamline the entire testing process,” Dr. Auerbach says. “We could now eliminate a major preanalytic step of aliquoting and the resultant labor, waste, potential for error, biohazard exposure, and resultant increase in turnaround time, and use one system for both our routine and stat testing.”

Excluding the cost of the analyzers, which Abington would have had to buy regardless of whether it automated, the additional cost of the integration, track, sample manager, and requisite software was about $400,000 to $500,000, he says.

Dr. Auerbach says having data managers as an interface between the automated analyzers and the laboratory information system has several advantages. Previously, the LIS would treat results from each of the analyzers as different methods. “But with a data manager,” he says, “the data from four different instruments are consolidated, and therefore reruns, reflex testing, and autovalidation can be managed more easily before results are passed to the LIS.”

The payoffs
Since Abington finished implementing automation in the first quarter of 2002, the core lab’s workload has continued to increase—it now serves 65 nursing homes and about 50 physician offices—while staffing has gone down. Automation has saved seven full-time equivalents through attrition, about 20 percent of the technical staff, Dr. Auerbach says.

In addition, worked hours per billable test declined from about 0.2 in 2000 to 0.15 in late 2002. The target was to be at least at the 35th percentile on the HBSI benchmarking data for hours per billable test, and by the third quarter of 2002 Abington had achieved that. Billable tests per FTE showed about a 20 percent improvement between late 2001 and late 2002. “We’ve climbed to 13,000 annual billable tests per FTE,” says Dr. Auerbach. “We can be semicompetitive with Quest,” the nearby reference laboratory.

Salary and supply costs and worked hours per core lab test have all declined. Multiple instruments can share one tube of a patient sample, reducing cost and the volume of biohazardous waste. The total direct cost of the average lab test, including labor and supplies, has dropped from $9.13 to $7.96, Durr reports.

The lab has also been able to consolidate its floor space usage. “We have most of our testing in a very confined area,” Dr. Auerbach says, though that required a new look at heat load and air-conditioning and similar things.

Turnaround times plummeted. For example, on the Advia WorkCell for chemistry analysis, 69 percent of tubes are sampled, archived, and sorted in the sample manager in less than six minutes. Stat results are available in 22.3 minutes and routine results in 24.1 minutes. (Those are median times.) Overall, the median stat turnaround dropped from 52 to 43 minutes after automation, while routine turnaround times dropped from 100 minutes to
61 minutes.

In chemistry, which represents three-fourths of the lab’s workload, 95 percent of test volume can be processed on one system, including 56 serum assays, 18 urine assays, and seven fluid assays. In hematology, 62 percent of CBC results are autovalidated, so that technologists spend their time managing only the exceptions.

The automation process can be managed from PCs within the laboratory. “We use keyboard video mouse technology, known as KVM switches, to control everything from PCs,” Dr. Auerbach says. Using a matrix type of KVM switch technology, the lab could integrate any analyzer that uses a PC or Unix workstation as a front-end controller, and thereby gain full control of the analyzer from multiple access points in the lab. And the new systems are robust. “Over 18 months of use we’ve seen less than 24 hours of downtime,” he says.

The glitches
One of the biggest problems in implementing automation was getting the staff accustomed to computer-controlled rather than personally controlled testing. “The technologists had to quickly ramp up their computer skills,” Dr. Auerbach says, to understand how integrated systems are networked, where the potential pitfalls are in the information flow, and how to restart the systems. Those systems now consisted of several computers, such as the data manager, the workflow manager, the LIS, and the actual analyzer computer.

“Automation requires a whole new skill set in your medical technologist,” Dr. Auerbach adds. “They were used to knowing exactly what they put on each instrument. Now they have to feed the automated workcells and keep a continuous flow going, and manage the exceptions. They have to think about integration.”

Medical technologists’ training in computer information systems has been limited, he notes. “They probably receive some training in their didactic curriculum, but in their practical year of training, their informatics experience will depend on the lab they trained at,” he says. They have to be prepared for a networked computer environment, and be able to work in the traditional role of medical laboratory scientist and to understand the informatics involved in an automation system.

“The technologist has to be able to recover from a computer lockup or system crash without having to call the help desk,” he says.

Under the nonautomated workflow, Durr adds, technologists and technicians would run one machine and do each test separately. With automation, “there was a different dynamic,” she says. “Now they had to share data and interact with other team members. They had to respond to the person who was running the data manager.”

Dr. Auerbach and Durr acknowledge that although upfront training was provided to the technologists and technicians, the lab underestimated the training needs, not just on the new equipment but also with new processes. Says Dr. Auerbach: “You can’t just buy a vendor’s solution and think it’s done. You have to work to get those FTE reductions. You have to figure out how to change your workflow.”

However, says Durr, “You really can’t experience it until you do it. At some point you have to take the plunge.”

The bottom line

All in all, lab automation at Abington Memorial Hospital has been a “resounding success,” Dr. Auerbach reports. “We’ve been able to reduce cost, achieve FTE reduction, and see opportunities for growth and additional revenue.” The managed care payers are now “coming to the table and discussing realistic contracts we can live with.”

Abington continues to tweak the automated systems. “We’ll be adding plasma as an option soon,” Dr. Auerbach says. Other projects include expanding autovalidation, consolidating the menu, refining how reagents and tubes are managed, stocking the system to make sure there’s enough of everything throughout the day, and offering computerized clinical alerts on important results via the hospital information system. “We want to expedite delivery of test results to physicians,” he says.

In summary, Dr. Auerbach recommends that any laboratory considering automation take the following steps:

  • Map out your current workflow and process.
  • Examine the entire process from specimen receipt to result.
  • Study the vendor options and their impact on your laboratory.
  • Consider the analyzers carefully because they are the potential bottlenecks.
  • Don’t overlook the data management/process control system because it is a major component of the system.
  • Consider contingency operations.
  • Start computer training for your technologists. “It’s never too soon,” he says.
Karen Southwick is a writer in San Francisco.