Feature Story

Order in the blood bank: automation steps up

October 2002
William Check, PhD

When David Wenham began working in blood banking more than 30 years ago, syphilis was the only infectious disease tested for. "I have seen a lot of changes since that time," says Wenham, who is a Fellow of the Institute of Biomedical Sciences and manager of processing, testing, and issue operations at the National Blood Service, Colindale Center (North London). Advances have come particularly in automated equipment, Wenham says. "In the early ’70s, we added testing for hepatitis B virus. It was purely manual at first. Then we moved to microtiter-based hemagglutination systems, on to RIAs, then to ELISAs." Microtiter-based EIAs appeared in the mid-’80s. Automation proceeded from plate readers to robotic sample processors, then gradually in the early ’90s to automated microplate processing. An ocean and a continent away, at the UCLA Medical Center, Loni Calhoun, MT(ASCP) SBB, senior technical specialist in the Division of Transfusion Medicine, describes a similar situation in antibody screening. Initially, it was done by a tube method. However, Calhoun says, "Our workload was growing and we couldn’t get more technologists. So we wanted a system we could automate."

Demand for automation is keen across all clinical laboratory fields. But it is particularly intense in blood banking. "Blood banking is one of the last bastions of true piecework in laboratory medicine," says Edward Lipford, MD, pathologist and medical director, Carolinas Laboratory Network. "We need to do anything we can to move blood banking into the current era of automation and greater efficiency while keeping the same highly specific patient results."

Dr. Lipford’s colleague, Jared Block, MD, medical director of the transfusion service of the Carolinas Laboratory Network and at Carolinas Medical Center in Charlotte, recently made site visits to look at new automated systems. "The justification I heard from many places was FTE savings," Dr. Block says. "We were down two FTEs. Other places were down many more."

Pressure for automation also derives from the greater number of bloodborne infectious diseases. In addition to a serologic test for Treponema pallidum, in the United States donor blood must be screened for the surface antigen of hepatitis B virus as well as for antibodies to HBV core, hepatitis C virus, HIV types 1 and 2, and human T cell lymphotropic virus types I and II. Also mandatory are tests for HCV and HIV RNA. Antibodies to cytomegalovirus may be sought in selected cases. And at its Sept. 12 meeting, the FDA’s Blood Products
Advisory Committee discussed adding tests for Chagas disease and West Nile virus.

One consequence of the drive to automation is described by Steve Negin, MT(ASCP)SBB, laboratory manager at the Central California Blood Center in Fresno. "It is getting to be almost impossible for small or even medium-sized donor centers to be cost-effective," Negin says. "Small centers are just not going to be able to afford automation. We have already started to see, as inspectors see more automated systems, they are asking smaller non-automated centers, ’How are you capturing the additional information that the automated system is capturing?’ Without automation, that is next to impossible."

Negin details some QC steps performed by the instrument that he uses to test for viral markers. In manual testing, the temperature of a reaction would be taken when putting in a plate and when taking it out. "This system continually monitors temperature in each slot," he says, "and it will lock out a slot if there is a problem." Moreover, the instrument profiles intermediate steps that weren’t even monitored before, such as exact wash times and times to add reagents, within seconds.

German Leparc, MD, chief medical officer of Florida Blood Services, St. Petersburg, agrees that automation provides benefits beyond staffing reduction. "Perhaps the biggest impact of automation has been in allowing us to meet current GMP requirements," Dr. Leparc says. "Such things as incubation times and other processing steps that before needed to be documented by hand are now done by the instrument. If limits are exceeded, the instrument will invalidate those results." No longer must technologists exhaustively review sample records when a problem is detected after a run.

Automated or semiautomated instruments are available for the two main blood bank functions—determining blood type and screening for antibodies (type and screen), and assaying for infectious diseases.

Type and screen

Newer automated or semiautomated instruments that type and screen timized for either donor collection settings (Immucor Rosys Plato) or patient transfusion settings (Immucor ABS 2000, Ortho Tecan MegaFlex). One exception is the Olympus Tango, which adapts to either setting but is available now only in Europe.

At Massachusetts General Hospital, Sunny Dzik, MD, co-director of the blood transfusion service, oversees testing of about 39,000 units of blood annually, either collected from donors or purchased from the Red Cross. "We use the Rosys principally to do a basic type and two-cell screen on blood donated at MGH," Dr. Dzik says. (Patient antibody screening is done on gel cards using a three-cell screen.) The Rosys can also perform syphilis screening and CMV serologic testing.

Dr. Dzik says the Rosys processes 94 samples per run and takes 1.5 hours to do the basic required set of tests. "It is definitely not a walkaway instrument," he says. For instance, a technologist is involved in physically transferring microtiter plates to a reader at a particular time. However, Dr. Dzik says, "I don’t see that as a major disadvantage."

In the Carolinas Laboratory Network, which services four hospitals in Mecklenburg County, the transfusion service transfuses about 35,000 units per year. They use both a Rosys and an ABS 2000 to perform type and screens. "These instruments allow us to do our work in large batches without using as much manpower," Dr. Block says. The Rosys does a two-cell screen, the ABS a four-cell screen. Says Barbara McElhiney, manager of transfusion services: "We do both donor and patient type and screens on the Rosys. The ABS 2000 is used primarily for patient type and screens, but it could be used for donor testing also."

In practice, Dr. Block explains, "We separate our workload. We do the majority of our work on the Rosys, which has higher throughput, while the ABS allows us to separate out smaller batches." Samples from obstetrical patients and trauma or surgical patients who are not bleeding heavily might be batched. A type and crossmatch used to take 45 minutes. "Now," Dr. Block says, "we can do four in under an hour, and they all come out at the same time."

He estimates that the instruments have saved 3.5 FTEs: "We were down two FTEs and the instruments allowed us to reduce staff by another 1.5 through attrition."

Dr. Block finds the instruments’ sensitivity to be good. "We have picked up some significant antibodies like anti-E and anti-Kell that we would not have picked up manually," he says. Antibodies were verified with sensitive manual techniques.

When they were evaluating the Ortho gel system, they were told that the solid-phase technology (Immucor) would produce a problematic level of false-positives. "Before purchasing the system," Dr. Block says, "we evaluated the presentation data prepared by an advocate of gel. We examined the data used to conclude that gel renders fewer false-positive results, and the data appeared to be flawed." The initial review found that the data sets were obtained from two separate populations of samples. Therefore, Dr. Block says, "the comparison was not truly apples to apples."

In addition, he says, the results within the studies on the gel system suggest that "significant antibodies were missed by this method—21 anti-E, seven Kidd, seven Kell, all picked up with PEG and not by gel." Dr. Block’s laboratory has seen about three false-positive results per month since it implemented the Immucor system in April of this year. "This is from an approximate sample size of 1,800 per month," he says.

The Immucor alternative had advantages. "The Ortho proposal placed a single instrument here, whereas the Immucor proposal placed both the Rosys and the smaller instrument," Dr. Lipford says. "That gave us what we thought was greater flexibility and some core redundancy in our automated blood banking operation for approximately the same initial capitalization."

Reagent cost was also a consideration, says Dr. Block, who recalls the Ortho system reagents as being far more expensive. "It actually amounted to a recurring reagent cost of $100,000 per year greater, even though the capital outlay was the same," Dr. Block says.

Another major factor, in Dr. Lipford’s view, was that the Immucor instruments are truly bar-coded, with an on-instrument bar-code reader. "Whereas the current version of the Ortho instrument reads the bar code first, then the instrument remembers its rack position. Since that is not direct tube reading, there is always a potential for error if you put the tube in the wrong place," Dr. Lipford notes.

Other institutions selected an Ortho instrument. In 1995, the transfusion service at Emory University Hospital, which transfuses 18,000 red cell units per year, moved from manual tube testing to a manual gel system, says Mary Beth Allen, MS, MT(ASCP), administrative director of laboratories. In 1998 they moved to full automation. "We do typing for patients and donors, antibody screens, and donor type confirmation," Allen says. "But our transfusion service is most important.

"We evaluated across the market," she continues, "including both Immucor and Ortho instruments." They purchased an Ortho Tecan MegaFlex. "It automates the entire pipetting process," Allen says. It also includes automated centrifuging and reading.

Since Emory is a tertiary care hospital, many patients have been transfused previously. "Typically we see 10 to 15 percent of patients with antibodies," Allen says. "With the Tecan we very rarely see a false-positive. With other systems we have seen some false-positive results."

The decision to go with automation at that particular time was driven entirely by staff reduction issues, Allen says: "Our staff was already reducing despite our best efforts." When they were doing manual tube testing, they had six FTEs on the day shift to do routine serological testing. When they moved to manual gels, it went down to two. Now only one person is required.

Allen is not disturbed that the Tecan is not a true walkaway, requiring an operator to transfer the gel card at various steps. "I don’t know that there is a true walkaway yet in the U.S.," she says. "Several are under development and may be available in Europe. But it can be a challenge to get these instruments through FDA; it’s a fairly rigorous process."

Allen’s advice: "Any facility that is considering moving to automation—which I heartily recommend—should evaluate everything on the market based on their specific facility."

At UCLA Medical Center, Calhoun helps oversee

technical operations in the transfusion service, which uses about 34,000 red blood cell products a year. They, too, chose the Ortho Tecan MegaFlex system to automate patient type and screens. Calling the MegaFlex their "workhorse," Calhoun says, "We tell it what tests we want to run on how many specimens; the instrument tells us what reagents and gel cards to add and where." Program prompts are easy to follow. "We load the specimens. It prepares cell suspensions, pipettes the cells and plasma, and tracks the identity of every test on every card by bar code," she says. Because the specimen bar code is not read until the sample is pipetted, incoming stats are easy to add—they simply substitute the next specimen to be bar-coded with the stat specimen. A batch of 24 specimens can be completed easily in an hour.

Although test cards must be manually moved for incubation and centrifugation, a companion centrifuge-reader simplifies result entry. The MTS Reader SA instrument centrifuges and reads up to 24 cards at one time. A picture of each card’s results, with strength of reactivity and result interpretation, is displayed on a computer screen. A technologist reviews the display and accepts or changes the data as needed, then exports the data to the LIS. "Because many of our patients have serologic discrepancies, we have chosen not to export interpretations at this time," Calhoun says. "A technologist reviews results in the laboratory computer for final interpretation."

Calhoun likes the Tecan system’s versatility. It can be set up to screen for or crossmatch donor units of specific phenotypes, such as when a patient needs E, c-negative blood. "We can use the pipetting instrument and reader for many test applications, as long as we properly validate them for that use," she says.

When making its choice, UCLA compared the Ortho gel and Immucor solid-phase methodologies in a parallel trial of 300 specimens. "Our primary selection criterion," Calhoun says, "was which method worked better in our setting." They were able to detect more clinically significant antibodies with gel than with solid phase. "Although solid phase was more sensitive, with our patient population it also detected many more autoantibodies of dubious significance, and these obscured the underlying alloantibodies we clearly detected with gel," she says. Gel reactions were consistently stronger than LISS-IgG tube reactions. Moreover, technologists participating in the trial study preferred the gel system and found it easier to adapt to.

Since implementing the Tecan, staff size has not changed, Calhoun says, but workload has increased and become more complex. "Automation has allowed us to re-engineer job tasks and use staff more efficiently. Plus we have a testing system in place that is objective, reliable, and reduces the risk of clerical error. Automation has saved us," she says.

The Olympus Tango, which is not available now in the U.S., promises even further advantages. (Biotest AG is the manufacturer of the Tango and associated reagents and distributes directly in Europe. In North America, Olympus will be the distributor.) While it can be difficult to compare instruments that have not been cleared by the FDA with those that have, interviews with laboratorians in the United Kingdom who are using this instrument suggest that it is accurate, flexible, and fully automated.

Queens Medical Centre, Nottingham, is a large acute-care teaching hospital with a transfusion service that performs about 30,000 blood groups per year and transfuses about 18,000 units of RBCs. "We do a full blood group and three-cell antibody screen," says Linda Hoyland, MSc, chief biomedical scientist and manager of the transfusion department. "This work is now totally automated on the Tango." She calls the instrument "a completely walkaway analyzer—you put the samples on, push a button, and get results."

In the summer of 1999, Hoyland and her colleagues compared four analyzers: Tango, DiaMed Diana gel station, Ortho CD Autoview, and Immucor ABS 2000. "We compared them on sample throughput, technologists employed, amount of skill needed, whether they were true walkaway, availability of 24-hour service, and serological performance," she says. Ultimately they chose the Tango.

The Tango’s variable throughput was an important factor. It can work in batch mode, with a capacity of 120 samples. In normal operation it takes 35 to 40 minutes to get results, according to Hoyland. But, she says, it is flexible and designed to be able to "trickle feed"—to add samples singly. For one sample, groups and screens take 35 minutes; eight samples take 40 minutes. "That is one of the things that helped us decide," she says. "Others claim you can put on one sample at a time, but may have restrictions. With some instruments, if you start a batch before the prior batch is finished, you can lose some of the earlier results."

She also likes the Tango’s serological performance. "I can’t fault it," she says. Its edit rate—results that the analyzer can’t interpret—is about four percent. "In practice, the actual number of samples that we have to work on manually is less than two percent," Hoyland says. She calls accuracy for antibody screening "very good." The false-positive rate is less than one percent.

No false-negative results have yet been recognized. In some patients whose historical records show an antibody three to four years ago, the machine did not detect it. But in these cases the antibody wasn’t confimed by sensitive manual testing either.

The Tango can do crossmatching, Hoyland says, and she plans to implement it when time allows.

In the Haematology and Blood Transfusion Department at Southampton General Hospital, a comparative evaluation also led to selection of the Tango, says laboratory manager Steve Yates, Fellow of the Institute of Biomedical Sciences. This service provides blood for hospital patients, performing 60,000 group and screens annually.

In the mid-’90s, the service had a semiautomated liquid-handling device that used microplate technology, but it had several limitations. "We needed someone looking after the machine or around it most of the time," Yates says. "We also felt unhappy about quality aspects of that device." Results varied with operator skill. Also, it could be used only for batch analysis, the minimum batch was 24 samples, and it was unsuitable for 24-hour use at their staffing level.

They compared approximately the same four instruments that Hoyland compared. How well each instrument performed serology was the key consideration. Having all processing done by the instrument itself was also important. "The Tango fit our needs best," Yates says. "It was the most sophisticated instrument on the market at that time," the end of 1999. "It still looks good today."

Serological performance is good. Yates estimates the false-positive rate for antibodies at less than 0.01 percent. For sensitivity, he says, "with serology that is always a fuzzy concept. We now regularly detect clinically significant antibodies that we cannot detect by a manual column agglutination technique." In the first six months of this year, the Tango detected the following antibody specificities that could not be detected by a standard antiglobulin technique and required enzyme-treated cells for verification: anti-D-14, anti-Jka-3, anti-Jkb-1, anti-c-1, anti-e-1, and anti-K-1.

Because the Tango is a true walkaway analyzer, the only substantive technologist time is loading of reagents, which tends to happen throughout the day. "We have gained at least one staff person," Yates says, noting that the lab was automated before Tango was introduced.

Tango’s microplate strip format has eight wells, which allows two anti-Ds, a true reagent control, and a check of reverse group. With the column resolution technique, there was a card with six wells for either antigen or plasma testing. "We felt that was too few to perform a safe group," Yates says, "because it allows only one well containing anti-D, no true reagent control, and no check of reverse group."

Olympus is in the process of scheduling a meeting with the FDA to review regulatory strategy and field trial protocols and hopes to start field trials late this year.

Also in Europe, Immucor is marketing a new automated system called Galileo. It is a fully automated, high-throughput, random-access blood bank system that performs patient and donor type and Rh, weak D testing, antibody screening (pooled cell, two-cell, and four-cell), antibody identification, CMV detection, syphilis detection, and crossmatching. Immucor is preparing for clinical trials in the United States.

Donor infectious disease testing

The Ortho Summit System, or OSS, and the Abbott Prism perform infectious disease tests of donor blood. Abbott’s Prism was FDA cleared in 1999, but the Prism assay approvals have been delayed because of the FDA’s compliance hold on Abbott’s manufacturing facilities.

At Florida Blood Services, Dr. Leparc says, "We do donor screening for ourselves as well as for other blood centers." Their annual workload adds up to almost half a million units. (They do just nucleic acid testing on another 500,000 units.)

Dr. Leparc’s test menu shows the variety of instruments that a large blood center must have. For blood typing and syphilis testing, they use an older instrument, an Olympus PK7200. They use another older, semiautomated instrument, the Immucor Dias system, to do antibody screens on blood donors. For HCV and HIV nucleic acid testing, they use Chiron/Gen-Probe FDA-licensed kits, which are based on transcription-mediated amplification, or TMA, technology.

For all ELISA testing for transmissible disease markers, they use the OSS. (Sometimes the Ortho Summit System is referred to as the Ortho Summit Processor; the Processor is the main component of the OSS.) The OSS can perform six ID tests for donor screening: HBV surface antigen and antibodies to HBV core, HCV, HIV, HTLV, and CMV. Dr. Leparc finds that the OSS has reduced his technologist requirement for this facet of testing by about 25 percent.

On the transfusion side, which Dr. Leparc calls "one of the least automated aspects of our operation," he uses the Ortho manual gel card system for blood typing and antibody screening, rather than Ortho’s semiautomated Tecan system. "The Tecan system works well for batching," Dr. Leparc says, "but it does not give us the fast turnaround times we need for emergencies." He considers the Immucor ABS 2000 "a fairly good automated instrument for the transfusion service," but is currently inhibited by its cost.

Another large blood center, Blood Systems Laboratories in Tempe, Ariz., has also adopted the OSS for infectious disease testing. Blood Systems Laboratories (BSL) is a division of Blood Systems Inc. (BSI) and does donor screening on 1.2 million units per year for both BSI and several other blood centers and hospitals across the country. A second BSL donor screening laboratory in Bedford, Tex., screens an additional 800,000 units.

Gene Robertson, PhD, director of operations for BSL/Tempe, describes the various systems the facility uses. Like Florida Blood Services, they use Chiron/Gen-Probe’s semiautomated TMA assay to do nucleic acid testing for HCV and HIV, and the Olympus PK7200 for blood group and syphilis testing. For antibody screening, Dr. Robertson uses the older, semiautomated Immucor Capture-R Ready Screen method. While he could theoretically transfer these assays to a Rosys, he says the main issue is throughput capability.

"As far as we understand, the volume we process would not be practical even on many Rosys systems," Dr. Robertson says. "We process an average of 3,000 samples per day, up to a maximum of 6,000."

During the national crisis after Sept. 11, the laboratory was overloaded, which allowed Dr. Robertson to calculate that the laboratory’s maximum capacity in a 24-hour period is 8,000 samples.

Donor infectious diesase tests are performed on the Ortho Summit System. "It is relatively new," Dr. Robertson says. "We have been on it for over two years. I would still consider it semiautomated, but it is much more automated than other systems that are available."

The OSS consists of two main instruments: A pipetting system, in which a robotic pipetting arm aspirates samples and dispenses samples and reagents into microplates. A processor that adds reagents to the microplates and incubates, washes, A technologist must first transfer the plates from the pipetting system to the processing system singly as they are prepared. Also, a technologist must add reagents to the processor or change reagents at certain times during the run as processing moves along. "A work list generator gives the technologist the tasks, so they have to attend to the system," Dr. Robertson says. "But the technologist is not performing any critical steps in the assays."

Another advantage is that all steps are documented by the system. "One of the main benefits we have realized is that the OSS documents GMP criteria," Dr. Robertson says. "It provides us with a complete run record."

Microplates have 96 wells, but with various controls each plate usually processes 88 samples. Calculating throughput or staffing efficiency in this large facility is complicated. The laboratory has eight processors and 14 pipetters, with one assay typically being done on each processor and one person "committed" to each processor. "We have two main processing shifts," Dr. Robertson says, "because the entire process cycle time is so long—
approximately 14 hours." Cleanup is done on the first shift. Within this framework, infectious disease tests on 3,000 to 6,000 samples are done each day.

Overall, Dr. Robertson estimates that the OSS has reduced staffing requirements for ID testing by about 25 percent.

He has tested the Abbott Prism. "Our laboratory did some clinical work on it," he says. "That is really a fully automated, truly walkaway technology. We had a very positive experience with the system and the results on the assays we performed."

Central California Blood Center in Fresno, with a testing volume of about 105,000 units per year, is a third large organization that is doing donor infectious disease testing on the OSS. "We are an independent blood center serving 30 hospitals in a five-county area," says laboratory manager Negin. "We also do testing for two other blood centers and two hospital-based blood centers." In June 1999, Central California Blood Center became the first blood center worldwide to go live with all six viral marker tests on the OSS, Negin says.

He notes that, in addition to the pipetter and processor, the OSS includes the capability to send raw data back to a built-in server running a proprietary program, Ortho Assay Software (OAS), which makes the final calculations on test results and interpretation and puts out results.

Internal QC features on the OSS score high with Negin—for example, the continual monitoring of the temperature of incubation slots and kicking out a plate and locking up a slot when the temperature goes out of range. "With this approach, you lose one plate but not the whole run," he says. Also, when dispensing reagents, the OSS checks to have a certain residual volume in the syringe, so it can detect mispipetting. "That can be hard to see with multi-channel pipettes," Negin says.

"And on the new software, if you don’t do weekly or monthly maintenance, it won’t let you start the machine," he adds.

Negin, like others, finds that automation has allowed him to increase workload without adding personnel. "On Sept. 1 of this year, we increased our volume by about 30 to 35 percent with no increase in FTEs," he says. "With our current configuration we could probably go to 115,000 to 120,000 samples before we needed more equipment or more people."

Experience with the Abbott Prism in England suggests that it may further improve donor infectious disease testing. At the National Blood Service’s Colindale Center, about 1,000 samples are processed each day, Wenham says. On the Prism, all donor units are tested for HBV surface antigen and for antibody to HCV and HIV. About 30 percent selectively get anti-CMV testing based on donor history. Anti-HTLV and anti-HBV core are not mandated in the UK. "In the U.S. you will use all the available channels," Wenham says, "so it will be even more cost-effective."

He says that the Prism’s advantage is that it is a fully integrated instrument: It pipettes all samples and reagents, processes plates, and does plate reading and data interpretation. It also sends data to a mainframe. "It is a completely closed system," Wenham says. "There is very little, if any, chance for an operator at any stage to interfere with either sample processing or plate processing or reading. So for safety and security, its GMP features are a long way ahead of open modular systems."

For QC, the Prism performs electronic tracking of samples, keeps event logs and error messages, and produces proofs of correct processing of each sample. "It is very cost-effective in terms of staffing, requiring fewer people than for the equivalent volume through microplate-based systems," Wenham says.

Throughput is high. Prism produces results for all assays processed at a rate of 150 samples per hour. Another welcome feature is constant access, making it possible to add 28-sample racks at any time without the need for batching. It can also process stat samples.

The Prism is also a walkaway instrument, with one limitation. Because Wenham’s laboratory routinely processes more samples than the Prism’s 280-sample capacity, a technologist is assigned to load and unload sample racks. But Wenham doesn’t see this occasional attendance as a major problem. "We don’t tend to walk away for long periods," he says. "There may be things we want to respond to, particularly the audible alarm that accompanies an error message on the screen."

To run ID tests on 1,000 samples per day, the laboratory has two Prisms running two shifts (and about 30 percent spare capacity). On each shift are two junior staff, mostly handling initial sample preparation, one "qualified" scientist running the instruments, and a supervisor who primarily oversees the Prism area. "This represents a reduction of two junior and one qualified person to perform this testing in comparison to our previous microplate systems," Wenham says.

Blood bank directors are optimistic about future improvements. "Automation in the U.S. is first-generation technology compared to chemistry," Allen says. "We are still in our infancy. True walkaway and random-access instruments will become the norm for blood banking," she predicts. "And prices will decline."

Dr. Lipford’s goal for automation is specific and ambitious. "So far vendors have automated some of the high-volume, highly repetitive aspects," he says. This includes the ability to run a limited antibody screen. "What is currently lacking," he says, "is a good system for antibody identification. That’s something we would be very interested in having."

William Check is a medical writer in Wilmette, Ill