Specimen speedway: How twin towers and a bridge keep hematology automation on track at Permanente
It looks a bit like the Tower of London at first glance. But the snaking, S-shaped
hematology automation system at the Permanente Medical Group Regional Laboratory
in Richmond, Calif., handles far more than royal blood.
Designed by Sysmex, a Japanese laboratory automation systems company, the three-line system is connected with a seven-foot-high bridge at the first bend of the "S" to enable greater throughput on a smaller footprint. The high bridge that gives it the towering appearance is necessary for ease of preventive maintenance and testing on the interior of the first two lines, says Richard Leung, assistant manager of hematology for Permanente.
The "S" shape was necessary to fit Permanente’s configuration of 12 hematology instruments into a newly built facility that’s part of a regional lab serving 3.2 million Permanente plan members in northern California. The facility handles about 70 percent of the lab work from a group of 18 hospitals and 35 medical centers, totaling about 20,000 specimens a day triaged into hematology, chemistry, and microbiology, says Gene Pawlick, MD, the laboratory’s medical director.
"We tried to get the latest and best technology we could use to most quickly and accurately process this large volume," which in the hematology area alone can reach 10,000 specimens daily, he says. "Because of the space constraints to lay this out, we couldn’t just run it in a linear fashion. We needed a smaller footprint."
Leung and Dr. Pawlick refrain from talking about the system’s cost, but they admit it was a multimillion-dollar investment. The first such installation in the United States, as well as Sysmex’s largest, the bridge setup at Permanente mirrors an earlier Sysmex arrangement in Japan, says Andy Hay, director of marketing for Sysmex America, Mundelein, Ill.
"The configuration with the tower was a piece [Permanente] could not have conceived of, but it already existed in a lab in Japan," he says. "To be able to turn it around and put it into a much smaller footprint made a big difference to them. They were struggling for space, even in a new building."
The system starts and ends with humans but has little need for them otherwise.
A laboratory assistant loads specimens onto a turntable at the lower bridge
on the opposite end of the "S," which divides them evenly between the first
two lines. When samples reach the bridge module, elevators carry them up one
tower, across the bridge, and down the other tower.
After undergoing a complete blood count and auto differential on one of the first two lines, samples that generate unusual results continue across another, lower bridge to the third line, which also performs a reticulocyte measurement if ordered.
Clinical laboratory scientists and lab assistants need not intervene, except to change reagents. Sysmex’s work area management system, called MOLIS WAM, makes the determination for them and generates an order for a smear. The third line also reconfirms the unusual results, Leung says. In addition to performing real-time data checking, MOLIS WAM stores up to 2 million patient results, he adds.
MOLIS WAM can process an even higher volume than the bridge setup alone would have, since technologists need not intercede, Hay says. "What we needed to do was introduce a layer of IT [information technology] below the LIS and above their instrumentation," he says. "It was kind of a ’Eureka’ moment when we sat down and discussed it with them. It now became clear that there was a good solution that would meet both their capacity requirements and their floor space constraints."
The system’s artificial intelligence makes determinations about whether additional testing is required based on our own prewritten rules, Dr. Pawlick says. "It’s just done if certain flags or parameters are hit," he says. If they are, "the artificial intelligence makes sure the specimen is conveyed onto our third track."
He adds: "Once you put a specimen on the loading end, the technologist doesn’t really do anything until all the testing is completed. Then the technologist verifies the results."
Neither the S-shaped tower configuration nor the MOLIS WAM software is necessary to process 10,000 specimens per day, Hay says, but without it, a lab would need a much larger footprint and considerably more staff.
"The key differentiator is exactly that: Nobody else has ever built anything that consolidated that is capable of producing that volume of work," he says. "If you can automate in this fashion, the number of operators is not linked to the number of instruments. There is no other single consolidated system that can even come close to matching this throughput."
The new system performs about 143 tests per instrument per hour, says Leung, meaning that with eight of the 12 instruments devoted to the basic CBC and differential testing, it can produce as many as 1,100 tests per hour. A single test takes from three to 12 minutes, depending on the need for the third line, he adds. Permanente’s old system, also designed by Sysmex, performed 120 tests per instrument per hour, Leung says.
The system is staffed with one clinical laboratory scientist and one laboratory
assistant most of the time, Leung says, with additional assistants brought in
during busy periods. "The work flow is totally different from the previous system
because we do not need human intervention," he says. "It creates an almost paperless
Previously, Permanente had an additional two staff members on hand per shift, Leung says, because abnormal specimens had to be found, taken from their rack, and manually placed on another rack for repeat testing. Then the other rack had to be told what test to perform, and the specimens eventually had to be put back on their original rack. "We totally eliminated that," he notes.
Those two additional staff members were reassigned to other areas of the lab, Dr. Pawlick says. He emphasizes that Permanente has a labor-management partnership that ensures "nobody loses their jobs due to technology."
Sysmex provided intensive training to bring Permanente’s staff up to speed on the system, Hay adds, putting 12 key operators through a weeklong class over a nine-month period that covered training on how to use the new technology. "They were already familiar with how to run a big HST [hematology sample transportation] system," he says. "We did a fair amount of on-site training for the people who hadn’t been able to come to our class."
Sysmex also ran half-day train-the-trainer sessions for about 20 people to help key operators develop their own training sessions for nonspecialized staff who sometimes rotate into the hematology departments for periods of time.
While specimens are loaded onto this system in much the same manner as on Permanente’s previous system, the use of the MOLIS WAM software required explanation, Leung says. He remembers learning how the work area manager performs, how to search for a particular specimen, how to find out the status of a particular specimen, and how to send out a MOLIS WAM-generated result to the wider laboratory information system. "MOLIS WAM is a middle-man before LIS," he explains.
The system, which went online April 26, has come a long way
from its crude schematic origins, Hay says. "We had several iterations of this,"
he notes. "The initial sketch was done on the back of a napkin at a restaurant.
We moved the configuration around in various ways," and the setup remains quite
flexible. "Although it looks like a big, permanent installation, it’s flexible
for potential future changes."
Sysmex builds systems like Permanente’s from standard modules, but the company fits installations to customer specifications in all cases, Hay says. "We recognize one system does not fit all," he notes. "We have also been watching the way the industry has been consolidating workstations."
Permanente wanted multiple lines of instrumentation in long straight lines, Hay recalls. "Dr. Pawlick had started with a vision of bringing a lot more testing in and making the organization more efficient," he says. "They were looking at three long, straight lines each capable of doing about 3,000 specimens" per day.
The entire system was not built in Japan, but enough of it was roughed out to be able to test the bridge and the computer communication system in that country. "The unusual piece of the mechanics was built overseas," Hay says.
Permanente staff traveled to Japan in September 2003 to see the prototype of the bridge working and to go over the technical details, so when Sysmex shipped the bridge to the United States, it would be ready to install, Dr. Pawlick says. "We saw the instruments, and they actually ran dummy specimens through them."
For testing purposes, Permanente had built a miniature replica of the system a year before it went live with the real system, Hay says. "They were quite lucky in that respect. Most of the hard work and most of the pain they went through was done in a testing environment," he notes. In that offline testing environment, he adds, there were many "hiccups," which were solved before installation.
That’s not to say the system has been glitch-free since then. Permanente has experienced some network speed and connectivity issues, which Hay terms "just part of the growing pains of a new building and adding instruments and services." He adds, "They initially had situations where they did more testing than they would have liked due to the network communication problems." Dr. Pawlick praises Sysmex for having "jumped in right away to solve these issues."
Leung agrees that communication and timing issues have been the main bugaboo thus far. "There are a lot of things involved: how the result is uploaded, how the order is downloaded-the abnormal reflex test order becomes uploading and downloading confirmation," he says. "The timing is very critical. It cannot be too fast or too slow."
Leung recalls that Sysmex sent a team of engineers from Japan to assemble and adjust the line. "They performed some function checks on the automated line to make sure the specimens were moving accordingly, without any stoppage." Since that team left, he says, local engineers have "fine-tuned it for calibration and revalidated that the movement of the hardware is functioning as it should be."
Ed Finkel is a writer in Evanston, Ill.