Feature Story

Blood banks join the automation chorus

October 2004
Anne Paxton

Heartland Blood Center, a mid-size facility in Aurora, Ill., was at a turning point in 1997: It needed to expand, it was running out of space, and it had to make a decision about its future. "We knew we couldn’t stay manual," says technical director Kathie Clark, BS, MT(ASCP)SBB. "Any blood center our size—or larger—has to be automated."

So when Ortho-Clinical Diagnostics approached Heartland about beta-testing a new automated donor screening testing system called Ortho Summit Processor (OSP, part of the Ortho Summit System), Heartland was ready to make the leap. When the system went live, however, it was no cakewalk for blood center staff.

In fact, "It was a bit of a nightmare," recalls Carol Genis, MS (ASCP) SBB, assistant technical director and education coordinator. "Admittedly, the techs begged to go back to manual testing at the beginning. There were several problems along the way, and there’s only so much you can train for."

But six months later, the learning curve had leveled out, the blood center’s workflow was successfully modified, and failures or voided plates dropped dramatically. Today, Heartland has two OSPs, and the medical technologists have done a 180-degree flip in attitude: "Now they wouldn’t give them up," Genis says.

And neither would management. Since the OSP was approved by the Food and Drug Administration, Heartland has grown from testing fewer than 100,000 specimens a year to 160,000 this year and a predicted 190,000 in 2005—nearly doubling its volume. "We’ll add a washer module to handle the increased volume next year, and maybe think later about adding another whole unit. But right now, the OSP has already allowed us to grow this much," Clark says.

Nationwide, the need to accommodate growth and the drive to boost efficiency are priming blood centers to acquire fully automated walkaway systems, either to perform types and screens or to test specimens for infectious diseases—or both.

In the U.S., four manufacturers of newer blood bank automation systems are vying for blood banks’ capital budget dollars. In infectious disease screening, the vendors are Abbott (Prism) and Ortho-Clinical Diagnostics (OSP). For types and screens, Ortho (ProVue), Immucor (ABS2000 and Galileo), and Olympus (Tango, pending FDA approval) offer fully automated systems. Olympus markets the PK7200 now.

Walkaway testing for bloodborne pathogens

"Automation has been late in coming to blood centers," says D. Michael Strong, PhD, executive vice president of operations at Puget Sound Blood Center, Seattle. One factor affecting the push to automate is that blood bank laboratories consolidated greatly because of nucleic acid-amplification testing during its clinical trials between 1999 and 2002.

"To achieve an IND [investigational new drug status] for the entire country, NAT testing had to be centralized to a limited number of laboratories, so many of the smaller labs had to partner with larger ones to provide samples for testing," Dr. Strong says. As a result, Puget Sound Blood Center tests about 439,000 units per year, with roughly 215,000 from its own center and the rest from others in Alaska, Hawaii, Washington, and Oregon. Now some mid-size blood banks are decentralizing, "and as automation is implemented it will make it easier," Dr. Strong says.

Puget Sound Blood Center has an unusual structure: It serves as both the regional donor center and the transfusion service for Seattle-area hospitals, most of which don’t have their own blood banks. Nevertheless, it exemplifies a trend at blood centers throughout the country: It is just starting to take full advantage of walkaway automation.

Automation on several fronts is online or in the works at Puget Sound Blood Center. For infectious disease testing, the blood center has two Abbott Diagnostics’ Prisms which it expects will be in operation by the end of the year. Abbott’s difficulties with the FDA have delayed licensure of the Prism reagents.

However, the Prism itself, a high-volume screening system that includes multiple channels to accommodate individual screening assays at up to 960 tests per hour, has been 510(k) cleared and is already used in several countries outside the United States.

"For ABO and Rh typing in the donor testing area, we use the automated Olympus PK7200, and Abbott’s semiautomated Commander System. But these are waiting to be replaced by the Prisms," Dr. Strong says. Prism includes assays for HIV-1/2 antibodies, hepatitis B surface antigen, hepatitis B core antibody, hepatitis C, and HTLV-1/-II.

Abbott signed a contract in January with America’s Blood Centers to supply Prisms to ABC members once the FDA has given approval for Abbott to market the instrument.

In Quebec, Prism has already helped the blood service achieve economies in staffing. France Bernier, MSC, is director of product qualifications for Héma-Québec, the blood service for the entire province of Quebec. Héma-Québec installed Abbott’s Prism in May 2002 and now has two units. Canadian Blood Services, which handles the rest of Canada, has Prisms throughout the country as well.

"We were working with EIA, we were testing for HBsAg, HCV, HIV, HTLV, and p24 antigen, so we had one technician per test," Bernier says. "When we introduced the Prism, it was able to do all of these tests except the p24, so we went from a total of eight technicians on the night and day shift to four." They also had a technician on the evening shift who was maintaining all the equipment, and they eliminated that position because there was less maintenance to do.

But Bernier says the greatest improvement was eliminating the human mistake. "Because when you do EIA you have to document everything—the reagents you are mixing, the equipment you’re using, the expiration dates of everything you use. So you have many things to write, and the more you write the more you can make mistakes," she says.

"With Prism the recording is very low," she adds. "In the past if somebody was doing the test correctly but forgot to write something, then we’d have to track the batch; we’d have to retest everything." That doesn’t happen with Prism, she says.

Speed is also greatly improved. "Once we put a sample on the machine, I think we get results in two or three hours. Before, it was almost five hours after the tests began. And when doing EIA you’d have to wait to have 500 specimens in the laboratory, because you’re processing six microplates by the technician. But with Prism you can put 100 samples on the machine, and if you receive 50 samples after that, you can put them on too; it’s ongoing."

Héma-Québec actually needs only one Prism to conduct all of its testing. "The second is a backup," Bernier explains. "We use both units, but not at the same time."

Ortho’s OSP, designed for mid- to high-volume screening, is the principal automated system available in the United States. It fully automates six blood donor screening assays—HCV version 3.0 ELISA, HIV-1 p24 antigen ELISA, HIV-1/HIV-2 peptide EIA , HTLV-1/II ELISA , HBsAg system 3.0, and HBc ELISA system—and new tests can be added as required.

Like Héma-Québec, Heartland Blood Center has found that one of the most helpful features of its new system is being able to modify a worklist while a batch of specimens is running. "If clients’ samples were coming in late due to bad weather or other circumstances, we couldn’t add onto the machine," Clark explains. "Until the software upgrade in 2002, you’d set up a worklist and you were committed to follow through to the end. Any additions would have to wait."

However, Heartland continues to conduct manual testing as well. "We are one of the few sites that does," Clark notes. "All new techs are required to learn all new manual tests, and we intentionally bring some testing back to manual format until they are fully competent."

"We also use it because it’s more convenient sometimes to throw some plates on manual. If there’s an error or a failure late in the evening, it’s more convenient to retest it manually rather than try to put it back on the OSP."

Clark says assays run between 90 minutes and 3.5 hours, for an individual plate with 88 samples.

The quest for efficiencies

Donor centers and transfusion services can benefit from the twin benefits of automation: reducing error and optimizing efficiencies, says Susan F. South, M.A.O.M., MT(ASCP), a health care consultant with Ortho-Clinical Diagnostics. An expert on "lean manufacturing" concepts, South frequently addresses transfusion services interested in automating types and screens.

"Blood centers are ripe for automation, in that a lot of what’s being done in many centers is very manual. There’s a lot of touch points for the operator, a lot of places where the operator is intervening, and that means more chance for error."

Perhaps surprisingly, it’s not saving time that’s considered the chief benefit of automation, but reducing the number of process steps.

Many vendors will say their instruments give high throughput, "but throughput isn’t what’s most important," South says. Understanding their process can spare blood banks from choosing automation they don’t need. "It doesn’t matter how good a piece of automation you select if you don’t understand how to integrate it into the process. It won’t work; it’s a waste of money," she adds.

One of the powerful aspects of using lean manufacturing concepts, she says, is being able to integrate materials, machines, physical space, and people. "And the ’people’ piece is always the most difficult in making change happen in any size facility."

Particularly in transfusion services, automation is seen as a foreign concept. "People think they are losing control, and becoming more like chemists," South says. But in fact, they don’t appreciate how much error they inject into the final result—what she calls "defect opportunities."

You can’t talk about automation and not talk about labor, she says. "Almost every administrator I’ve talked with is looking at decreasing labor. Whether they don’t have enough, or the cost is too high, or there’s too much volume for the available staff, it all comes down to labor. For every one bringing in automation for quality reasons, there are 100 bringing it in for labor issues."

The vast majority of transfusion services are in a semiautomated mode, she adds, but those that have gone to the fully automated systems have justified the purchase by saying they need to cut labor costs.

Lean manufacturing concepts are new to health care and to laboratories, South says. They’re about "fixing the process from the inside out, and there’s no end to the application of these principles," South says.

"With a lot of projects, it will be a year before you see results and they will amount to 10 percent, whereas the average I’m seeing is 75 to 80 percent reductions in cycle time, a 50 percent increase in productivity, a 50 to 90 percent decrease in nonproductive inventory, a 30 to 50 percent reduction in the amount of space needed to do work, a 75 to 80 percent reduction in the distance operators have to travel to do work, and a minimum 10 percent reduction in defects."

Using failure mode and effects analysis and value-added time and defect opportunity analysis, South and her colleagues have drawn dramatic contrasts between manual tube and manual gel methods of doing types and screens, versus using automated testing.

Although cycle time did not differ significantly, manual tube testing was found to employ 27 process steps with 173 defect opportunities, while manual gel had 12 process steps with 75 defect opportunities—and Ortho’s ProVue had only three process steps with only 16 defect opportunities.

Walkaway types and screens

"Automation in general is a very big topic at most meetings," reports Linda Cardine, MT(ASCP) SBB, blood bank supervisor for Henry Ford Hospital, Detroit. "Most of us realize blood banking is so far behind. I don’t want to say we’re the last to become automated—but we’re still on tube testing."

On June 14, however, Henry Ford became the first blood bank in the U.S. to acquire Immucor’s Galileo, a second-generation automated system.

Cardine had heard about Galileo when it was available in Europe but hadn’t yet been approved here. "Europe looks at laboratory testing a little differently. They’re much more laid-back, and their stat is not what we call a stat. We Americans think we should have results yesterday, so I’m not always sure the European experience can cross over very completely. But we’re going to use the instrument very hard."

The reagents cost three times as much as those for tube testing, she admits. "But this was a capital expenditure the laboratory was willing to make," she says. It was what the blood bank had to do to get the work done, a situation that parallels what clinical laboratories went through decades ago.

"Somebody had to buy that first Coulter for hematology," she points out. "The hard part is not replacing staff when you’re justifying the expense. But the staff are very supportive of the fact that if we don’t give the technology a try, we’ll fall behind."

"We had a shortage of techs, while our workload increased 20 percent last year. Henry Ford is a very robust health system with a level-1 trauma center, and another transplant team was added this year." The transfusion service does just over 4,000 types and screens a month. So if the technologists were busy testing, it was hard to answer the phones and get products out to customers. "What was suffering was customer service," Cardine says.

Given the expense of automating, Cardine wanted to get something like a chemistry instrument that technologists could load and walk away from, but also add to continuously.

On the Galileo, whether the runs have one specimen or 42, normal turnaround time is 45 minutes, she says. "So if you do have large batches, the throughput can be very quick." At the same time, she likes the linear tracks that make it possible to add and remove specimens easily. "It’s like a train track; as soon as there’s a green light out on the track, you can take the specimen off right away if you need to set up a quick test or do something else."

The Galileo footprint is manageable, too. "It’s about the same as a six-burner stove, if you do gourmet cooking. We’re building new countertops around ours, but at the moment it’s just self-contained. You plug it in and connect the phone line."

The blood bank at St. Joseph’s Hospital and Medical Center, Phoenix, has added several full-time equivalents since installing its Immucor ABS-2000 in February 2000, says supervisor Lawrence Rodriguez. But it’s mainly because automation has allowed the blood bank’s volume to nearly double.

In 1999, St Joseph’s was doing 14,000 types and screens per year. "There was a shortage of MTs, we couldn’t find people, we needed to do more with less, so we looked at automation," Rodriguez says. Now that the blood bank’s volume is almost at 25,000 types and screens, it needs an instrument with more throughput, so it is shifting to Immucor’s Galileo.

With the ABS2000, we "tweaked our workflow process," Rodriguez says. "We were running about four samples per run, and we would just keep it going 24 hours a day. It’s been a workhorse and has worked wonderfully for us, but now with the Galileo we can make runs of 10 or 12 at a time. And the continuous access will let us keep putting samples on."

The Galileo, he says, can put out 70 types and screens per hour using a four-cell screen, and would go faster with a two-cell screen. In addition to types and screens, the instrument performs weak Ds, antibody IDs, crossmatches, immediate spin and Coombs’, direct antiglobulin tests, cord bloods, and re-typing of units.

St. Joseph’s is in the middle of a $160 million expansion that will make it the largest hospital in the state, Rodriguez says. "Because of our expansion, we’re gearing up and we brought the instrument in to help us out. Automation is good, but it’s an aide—like having another technologist who doesn’t have anything else going on."

The technologists still have to thaw the plasma, issue platelets, accept blood, answer the phones, and perform numerous other functions, he notes.

"The technologist is always going to win if it’s a head-to-head, timed competition between the instrument and the tech, but where the instrument beats you out is standardization and not being interrupted by external influences. It just keeps moving along, whereas the tech can get bogged down with other stuff."

"With the ABS2000," he says, "we had a sensitivity of 95 or 96 percent and a specificity of 93 percent. We have not quite finished our Galileo parallel testing yet, but we have detected some antibodies on Galileo that did not get detected by the ABS2000." Of course, no one instrument can detect everything, he says. There are going to be icteric, hemolytic, and lipemic samples, and antibodies that may evade detection. "All these factors influence detection limits of automated instruments in blood banking," he says.

The interface on the ABS2000 is unidirectional, he notes. "But once you build your worksheet with your types and screens, it will drop in results for you, so it eliminates clerical errors when entering patient results."

Galileo’s interface is in development and will be bidirectional. "It will contribute as a time-saving measure," Rodriguez says, "in addition to eliminating clerical errors."

Error-free typing

When Immucor’s ABS2000 became available in 1998, the blood bank at State University of New York Hospital at Stonybrook was one of the first in the nation to step up to full automation, says Robert Borley, MS, MT(ASCP)SBB, associate technical director. It was a natural step forward because the blood bank was already semiautomated with solid-phase technology for antibody and cytomegalo virus testing.

After types and screens were added, "we realized about an 11 percent to 14 percent reduction in staffing needs for those laboratory tests, and we had been under a lot of pressure to expand to different testing that we didn’t have staff for."

Once they automated, they had about four FTEs they could then dedicate to bringing new testing online in other areas of the blood bank, including platelet testing and in-house leukoreduction of all their blood. "It actually was a major cost savings to the hospital," Borley says.

The blood bank moved to a new, larger facility in March 2003—but its laboratory space actually shrank by 25 percent. "We got less floor space, but because of automation you need less," Borley says. Donor operations and the apheresis areas were the beneficiaries of the added space. With a 4-foot by 2-foot countertop footprint, the ABS2000 was "not a huge factor in our space considerations," he adds.

Record keeping of quality control is now much simpler. "When you run automation, it becomes an automated process of the instrument, so it’s ’hands-off.’ You just run it similar to the way you run samples, and it puts a QC report out. It permanently attaches a QC report to every single result, which is all that’s required by regulatory agencies."

One of his favorite things about the ABS2000: "We have a history in excess of a quarter of a million typings without an error. With this system you have so many checks. You’ll get a refusal to type [NTD or no type determined], but not a wrong typing. No one would ever quote an error rate for manual typing, but I can tell you it’s high. And it doesn’t necessarily mean the person gets it wrong; it’s something you catch. But you never want to be in the situation of a near-miss."

With an NTD, the instrument still shows reactions, but not a defined type. "For example, if you were to run an infant, they are late in developing normal ABO antibodies, and usually you’ll find a negative serum type and a normal cell type. If the cell type does not match the serum type, you’ll get an NTD. It frequently happens also on bone marrow transplants because you’ll transplant a different blood type, so the cell type doesn’t match."

In manual testing, "you’d have to make technical changes for various disease conditions, and you have to do the same thing on automation after it gives an NTD." NTDs are given in less than one percent of the types, which reflects the proportion of a normal patient population that has these problems, he adds.

No monitoring is needed while the instrument is running. "It drops down, you push a button on the computer screen, you walk away, and it beeps when it’s done." Each instrument handles up to 16 samples in a batch, with the first four completed in 43 minutes, he says. The blood bank staggers its three instruments each with batches of four, so the instruments are running 12 samples at the same time.

Some technologists say they could manually do the same number in less time, he says. "Yes, you can. If you don’t answer any phones and don’t get up to go anywhere. But that’s not reality. The instrument is not influenced by phone calls or breaks, so it’s continuing to output every 43 minutes, and it far exceeds the capability and speed of what technologists can do overall."

Is there a tipping point at which a blood bank can decide it’s time to automate? In Borley’s view, that requires calculating the value of accuracy. "What is the value of having a tech who has spent two days in your blood bank coming straight out of technology school and can produce the quality of screen of a technologist of 18 years, because the instrument does it?" Automation doesn’t replace experienced blood bank technologists, he notes, but it does make it possible for new technologists to be introduced at a less demanding pace. "If a person with no experience can be trained in one day to produce 75 percent of your work, that’s a real advantage—especially in today’s labor market," he adds.

Puget Sound Blood Center has tested the Galileo as well as Olympus’ Tango and is evaluating whether to buy one for automated typing and screening. "That really is on the donor testing side," says Dr. Strong. "They have some advantages, but they’re not necessarily the same ones we might want in the blood transfusion services. We’re currently negotiating to purchase a separate system there from Ortho."

"The transfusion service side is maybe the last in the clinical laboratory arena to actually consider automation. It hasn’t changed much in the 60-some years of its existence, but now there are a lot of newer instruments."

Dawn Rumsey, ART (CSLT), who directs Puget Sound Blood Center’s centralized transfusion service, says it is the most difficult to automate. "There’s no predicting when our trauma center will have a car wreck or a gunshot wound. The numbers aren’t as stable as on donor testing, and that makes it difficult for manufacturers."

However, with emerging testing methods like plastic microtiter plates and plastic cards with microcolumns embedded, automation has become more practical, and Puget Sound has been evaluating the three available fully automated walkaway systems. It decided to go with Ortho and is negotiating a contract to buy two ProVues using gel technology, with plans for a third at a separate site.

"What we’ve found with ProVue is the software literally gives you a camera picture of the result that just doesn’t fit the instruments’ parameters, so you make the judgment call," Rumsey says. "It also puts cards aside for manual checks. The instrument has become much smarter—but control is still in the operator’s hands, not the instrument’s."

Small hospitals automate, too

Charles Ward, BS, MT(ASCP), is laboratory manager of the blood bank at Providence Centralia Hospital, Centralia, Wash., a small hospital with 110 acute-care beds and 80 extended-care or rehab beds that chose the Ortho ProVue a year ago.

"The reason we approached automation was that we probably do 6,000 or 7,000 crossmatches a year out of 15,000 billable tests. And the hospital’s oncology group was requiring more expertise than our generalist could handle. So we were looking for a tool to minimize the labor impact."

"We were doing gel technology manually and we basically had to be there through the whole process," he says. "When we went to automation, though, we could walk away. So we saw no increase in our labor needs and people were able to be used back at their specialty."

Moving from manual testing to gel technology cost the lab a little more, but the errors of mistyping were "drastically reduced," Ward says. "Some of those issues really did almost disappear."

Although the blood products had been bar-coded for more than 10 years, "bar codes were not being utilized very well in the blood bank because there was nothing over there except manual testing. When we went to ProVue, it was very helpful to have bar-coding." He says the advantage of ProVue is positive identification. "It’s a host-query bar-code system and asks which assays you want done."

Lab information system compatibility, he warns, can be a stumbling block. "Our LIS vendor does not have blood bank software, and at some point we expect to upgrade to Cerner’s Millennium software package. That will close the door on what we still do manually paperwise for issuing and tracking units of blood. We’d love to do it tomorrow, but there are about 1.3 million reasons not to do it right now," he adds, referring to the urgency of other priorities in the blood bank.

The LIS interface remains up in the air, too, at Henry Ford, even though the instrument itself has bidirectional interfaces. "It’s not the instrument’s fault, but because of the computer, unfortunately, we’re not at the moment connected to the LIS," Cardine says, citing incompatible coding.

"Right now we’re having to enter results by hand—though it’s no worse than when we were doing tube testing. It will get much more advanced when we get the interface going sometime in the next six to nine months," Cardine says. In the meantime, the LIS vendor is working with Immucor on an interim solution.

At Providence Centralia Hospital, the remaining paperwork bottleneck occurs around blood unit sign-outs by the nursing service. "To close that loop," Ward says, "we are looking at bar-coded wristbands. Lattice Software has a blood bank module where you actually download from the computer in real time, walk to the bedside, ’blink’ the wristband, it will say ’yes unit, here’s the bar-coded unit number that should be on,’ you bar-code the bag, and you can hang it." That should be in place in 2005 or 2006, he says.

"The thing we have found very interesting about ProVue is we see certain patients that probably, in years gone by, were mistyped. Especially the weak Ds, because of the technology we used, would have been reported out negative. But the gel technology is much more sensitive; it enhances those weak Ds and we’re now finding some we would have reported out negative five or 10 years ago." Some of the same is true in antibody detection, he adds. "So basically it’s gel technology that’s better than the old manual technique, but with ProVue we know sooner than we did before."

"We used to do a big batch on the evening shift. Now we process the workload as it comes in."

Any blood center that automates, he says, is probably going to find it a little more costly, "but the benefits of the workflow change can offset that, and there are people you might not have to add as you grow. On a given day, we used to have one to two FTEs working on the blood bank, and now we probably have one total for the whole three shifts, if you really counted the amount of time they actually spend in the blood bank. Because of automating, they’re able to walk away and do other things."

In Europe for the last three years and awaiting FDA approval in the U.S. is the Olympus Tango. Since 1992, Olympus has sold its high-volume PK7200 Automated Microplate System to large blood centers in the U.S., and the PK7200 in fact does ABO/Rh, syphilis, and CMV testing for most of the North American blood supply. But Tango (which also does blood group serology testing including antibody screens and has a custom test menu) is flexible enough to be placed in small to large transfusion services, donor centers, and reference labs.

The University of Iowa Hospitals and Clinics’ DeGowin Blood Center was one of six field trial sites in 2003. It performed parallel testing on Tango while it continued using its Immucor Rosys Plato, says clinical laboratory manager Judy Levitt, MT(ASCP)SBB. DeGowin Blood Center tests about 50,000 blood donor and patient samples a year, making it one of the largest hospital blood banks in Iowa.

"When we purchased Rosys Plato, there were no other comparable automated blood grouping analyzers for hospitals on the market," Levitt says. "Tango allows you to load samples on the instrument and walk away-which Rosys does to a lesser extent. But it requires hands-on manual transfer off the instrument and back on, whereas Tango is self-contained."

Tango offered inprovements in process control by monitoring items such as reagent shelf-life and storage and incubation temperatures. "These types of features assist with compliance," Levitt says. She says they were pleased with the Tango assay performance during the field trials.

DeGowin plans to request funding for the capital purchase of a new automated blood grouping analyzer next year. "We’re at the point where we feel the automation we have is dated and we need to replace it with current technology," she says.

Not a cure-all

Clinical laboratories have already found out that full automation is no panacea, notes Ortho’s South. Because they sometimes adopt automation reflexively—because the competition has it, or there’s a perceived need and not a real need—a few have had to go back and regroup. "Some places may spend a million or more on an automated track, they get it in, it goes down, and it actually slows down their cycle time. It’s not unusual for places like that to take out automation."

Blood centers, however, are in a position to reap much greater gains. Says St. Joseph’s Rodriguez, "It’s still early, and people have yet to quite embrace automation."

Anne Paxton is a writer in Seattle.