May 2011
Feature Story

Anne Paxton

Productivity. Reliability. Precision. Speed. Savings. The benefits of automation are well known in the clinical laboratory world, which, spurred by competition and shortages of skilled technologists, has long relied on machines to perform many phases of laboratory testing. But the automation that has worked so well in chemistry and hematology has been slow to take root in microbiology, traditionally regarded as too complex, for the most part, not to be performed manually.

Now, however, advances in robotics and innovations in laboratory methods are opening a new era. “I see us at the cusp of full laboratory automation in the microbiology lab,” says Nathan A. Ledeboer, PhD, D(ABMM), assistant professor of pathology, Medical College of Wisconsin, and medical director of clinical microbiology and molecular diagnostics, Dynacare Laboratories and Froedtert Hospital, Milwaukee. Indeed, it’s a trend that appears increasingly likely as many microbiology labs take the first step: the automating of preanalytics.

In the past three years, several automated specimen processing instruments have been placed on the European market, and two of them are now available in the United States. The WASP (Walk Away Specimen Processor), manufactured by Copan in Italy and distributed by Copan Diagnostics, Murrieta, Calif., was introduced in 2008 as a new system for automated plating and streaking of all microbiology samples. In 2010 the next-generation WASP became the first instrument in the world to allow laboratory personnel to load all samples onto a conveyor, then walk away while the machine scans the tubes and processes the specimens entirely, ultimately sending them out on an exit conveyor with the plates ready to incubate, says Copan executive vice president Norman Sharples. Copan advertises the WASP as having a throughput equivalent to two or three full-time people.

The second preanalytical device now marketed in the U.S. is the PREVI Isola, made by the French firm bioMérieux. Featuring a throughput of up to 180 plates per hour, the PREVI Isola employs advanced robotics to inoculate, streak, and label specimens, managing 90 percent of the steps required to process liquid microbiology specimen samples. The PREVI Isola’s comb technology inoculates the maximum agar surface, ensuring that optimum pressure-controlled contact with the agar is maintained and that a standard quantity of inoculum is used every time, bioMérieux says in the product literature for the instrument.

BioMérieux estimates it has about 100 installations worldwide; Copan says it has about 100 as well. Copan also offers an optional automated slide preparation instrument, the GramSlidePrep.

Another instrument, the Innova Preanalytical Automated Microbiology Specimen Processor, formerly made by Canadian company Dynacon, is now owned by Becton Dickinson but not currently on the U.S. market, though it is being marketed in Europe.

One other manufacturer is now selling similar instrumentation in Europe: Kiestra Lab Automation, which specializes in total lab automation for the bacteriology lab. Kiestra stresses the modular, open architecture of its system components, which can be adjusted to fit the space available. The company claims that a manual laboratory can double its productivity using Kiestra’s total lab automation approach.

Paul Bourbeau, PhD, director of microbiology laboratories for Geisinger Medical Laboratories, Danville, Pa., says his lab has been using Copan’s WASP since 2008. “We were the first lab outside of Italy with a WASP. But the instrument we have right now has evolved a lot since then. It’s gone from being a ‘beta’ device to a very mature, robust instrument that’s used every day in the lab and that can easily shift from urine specimens to stool specimens to ESwabs.”

Several factors converged to make preanalytic automation in microbiology increasingly necessary and increasingly feasible, according to Copan’s Sharples, a U.K.-trained microbiologist. A call for more screening for multiresistant bacteria has sent workloads soaring at a time when the supply of medical technologists is short and an entire generation is set to retire. “Microbiology desperately needs automation in specimen processing,” he says. Copan, which holds 95 percent of the U.S. market for specimen collection systems, already had the right background to fill the gap, but it first had to bring its large factory robotics down in size.

That need was met with Toshiba SCARA robots. (SCARA stands for Selective Compliant Assembly Robot Arm or Selective Compliant Articulated Robot Arm; Toshiba uses these interchangeably.) Like a human arm, the robot has a jointed, two-link layout that allows it to reach into confined areas and then fold back out of the way. WASP employs two of the SCARA robots to perform the tasks a medical technologist would normally do: opening the specimen container, dipping in a loop, planting the sample onto a plate, and streaking the plate using traditional quadrant streaking techniques. If so instructed, WASP can also de-cap a culture broth tube, inoculate a loop full of specimen into the broth, and re-close the cap.

The other advance that is paving the way for automation is standardization of specimens to liquids. Chemistry and hematology have always had the advantage of having liquid specimens, Sharples says. “But microbiology has a buffet of different types, and until now no one has been able to develop the robotics to take a swab out of a tube, roll it on a plate, and put it back. So we had to invent a new type of swab.” Copan’s ESwab, the only real liquid-based microbiology culture method, works by removing the sample, whether it’s liquid, solid, or semi-solid, immediately into a liquid. “So instead of being in a physical swab, the sample is now in suspension.”

The specimens that can be run on any of the automated systems must all be liquid, such as a urine specimen, a sputum specimen that was processed to make it liquefied, or an ESwab, Dr. Bourbeau says. Some laboratories, he notes, are accepting the added expense of ESwab, with the justification that they can offset the cost with labor savings from automated instrumentation.

For the inoculation method, Copan opted to use the standardized inoculation loop that has been around for 100 years, Sharples says. It’s versatile and can deal with all kinds of viscosity, and it gives appropriate volume transfers, which means better results. “The thing about the loop is you’re not re-inventing anything; you’re using the very traditional four-quadrant streak that every microbiologist used at school.” While Copan considered using combs, which create a spiral pattern, “to be honest they require re-education of the user.” Loops also give much better colonial isolation, he says. “With combs, it’s a bit like dipping a brush into paint. If there’s too much paint on the brush and you try to spread it, sometimes you don’t get isolated colonies. That’s one of the downsides of using the combs.”

When interfaced with a laboratory information system, the instruments make few errors—essentially no errors—unless something is done wrong in the setup, Dr. Bourbeau says. “So the idea of ‘Today we set up 200 urines and found cultures from 199 and we can’t find plates from the other one but we found two sets of plates on Mary Jones’—those things just don’t happen.”

The improved precision about time of inoculation is also helpful, Dr. Bourbeau adds. “By having that kind of check, you can be absolutely sure your plate has incubated for 18 hours, and that it did not come into the lab 18 hours ago and sit there for three hours before somebody plated it.”

While Geisinger tests 80,000 to 100,000 urine specimens a year in the microbiology lab, Dr. Bourbeau does not think the large-volume labs are the only market for the instruments. One year ago that might have seemed to be the case, he says, but on a recent trip to Italy, he found that relatively small labs are signing on to specimen processing automation. It’s partly due to the growing shortage of medical technologists in the lab. “The second thing is a lot of European labs are looking at a much broader picture: total laboratory automation. They’re saying, ‘Look, we have automation for chemistry and hematology, and the next step is automation for microbiology.’ So it’s not so much the number of specimens you do but the mindset of how you do it.”

Dr. Bourbeau expects culture-based methods to remain dominant in microbiology. “We started doing our first ‘molecular’ test in 1994 and our volumes have grown dramatically. We now do 100,000 molecular tests a year, and we use molecular testing rather than culture to screen patients for MRSA [methicillin-resistant Staphylococcus aureus]. But the percentage of total test volume that is what I would call core bacteriology is the same as 10 years ago, about 60 percent.”

The growth of molecular, he explains, has come from two things: culture-based methods that have been supplanted by molecular methods, and new types of testing that have basically gone straight to molecular. “So molecular is going to whittle away some testing a little around the edges, but I don’t see the proportion of bacteriology changing in the next five or 10 years, or even later. There are too many reasons we need to continue to do culture.”

The Dynacare Laboratories microbiology laboratory in Milwaukee has already taken a leap toward total lab automation. The first lab in the U.S. to report clinical results with the MALDI-TOF mass spectrometry system, it is now considering different systems of specimen processing automation.

One of the chief considerations is the types of specimens a laboratory receives, Dr. Ledeboer says. “What kind of containers do they come in, are they compatible with the automated instrument systems currently available, what level of manual processing is required with each of those systems? All of those are major issues you have to think about.”

Automated systems beat manual plating in some ways, Dr. Ledeboer says. “Certainly you get much more consistent plating, you get more consistent application of the specimen, and you’re going to get an easier read.” Admittedly, “Humans can get them done faster,” and on its own, automated specimen processing does not shorten turnaround time. “But automation will eventually lead to reduced turnaround time, because it will move us more toward the ‘Microbiology 24/7’ concept.” Most microbiology labs now do their work during the first shift, when the majority of specimens are read and reported. “The problem is, that leaves us with two-thirds of the day when we’re not reading cultures. So if we can move away from this pattern by reducing the human element needed to run the microbiology lab, we can start performing testing when the organisms are ready to be identified. For example, if we get a specimen at midnight, we don’t have to wait for essentially 36 hours before identifying it. We can test it maybe after 12 or 15 hours.”

The sterilized loops that Copan’s WASP and BD’s Innova systems use have the benefit that they don’t change the way the technologist reads the plate, Dr. Ledeboer says. “They allow us to see plates the way we’re used to seeing them, so we can do quantitative streaking of urines, of bronchial alveolar lavages, and so on.”

This also makes the use of bi-plates practical once more. “Historically, one of the limitations in the microbiology lab is the number of different media you can include in your algorithm. That’s been eliminated with the second generation of WASP’s inoculation system. If you have somebody trying to manually streak a urine onto a bi-plate and count colonies, and trying to manually identify isolated colonies, it can be a very significant challenge. When you automate, you make that process much more precise.”

For these reasons, he feels the automated loop inoculation systems may be favorable to technologists versus the combs technology used by bioMérieux’s PREVI Isola. On the other hand, because they are disposable, the combs are a little faster. “You inoculate the plate manually or have the instrument inoculate the plate from a decapped tube, then the instrument puts the comb on and the plate is rotated, so you have a circular pattern of organism and you get isolated colonies at the feathered edge. If you manually inoculate the plate, you don’t have the limitation on the number of media you can include in an inoculum. This is beneficial for stool cultures, or cystic fibrosis cultures, where you have maybe five, six, or seven pieces of media. With the comb approach, the system doesn’t care. However, the downside is you really have to train the technologist to look at the plate in a different manner,” Dr. Ledeboer says.

Automating specimen processing first makes sense, in his view. “It’s laborious and, from the technologist’s standpoint, kind of boring to do. It’s really an easy way to increase reproducibility and re-task the technologists to something they’re more interested in doing.”

A number of studies have found that automated instruments can reduce personnel costs. “The ability to generate savings with plating in a large lab will probably save one to two FTEs per shift,” Dr. Ledeboer says. Based on current prices of specimen processing instruments in the low six-figures, that savings would lead to a return on investment in about three years, he estimates. More impressive savings will come, of course, when the whole lab is automated, but larger labs with volumes of 3,000 to 5,000, or more, identifications a month would be the ones that could justify buying total line automation. “You may not have to be that large just to automate one process. Smaller labs may decide they just want to automate picking colonies or setting up of specimens, and they will have the choice of the pieces they want.”

He thinks pressure to automate the microbiology lab will lead to more consolidation, which could have unfortunate consequences. “The big players will be able to afford and justify purchasing automation, while the smaller labs aren’t going to be able to do so. And we’re really going to see more and more of the ‘have-nots,’ if you will.”

Like many large labs, North Hollywood-based SCPMG Regional Reference Laboratories, the Kaiser Permanente reference lab for southern California, has had automated plating and streaking for some time. “In our regional laboratory, we have had plating instruments for about 11 years,” says Susan M. Novak-Weekley, PhD, director of the microbiology laboratory. Her lab is now evaluating the BD Innova, which is expected to be back on the market soon, though instruments have been in place and are being used in Europe. Kaiser’s northern California laboratory has already opted to install Copan’s WASPs, whereas Kaiser’s Oregon laboratory is assessing the use of the Innova.

With prices for the newer instruments in the $180,000 to $250,000 range, it’s not a step taken lightly, but one that’s essential for a lab that does 1.3 million microbiology tests a year, “as our lab does,” Dr. Novak-Weekley says. “We know, because we plate so many specimens, that we have to automate; it was really kind of a no-brainer for us.” The seven Dynacon Inoculab instruments her lab currently uses were designed to plate tube specimens such as urine specimens and specimens such as enrichment broth for group B strep. The old Isoplater system from Vista Technologies has been useful historically but is being replaced by fully automated systems. “With the Isoplater the technician still has to take the specimen, label the plates being inoculated, uncap the specimen, place the sample on the plate, and stack the plate into the instrument, which then automates the spiral streak.” Her lab is still using six of those instruments, but it’s time to move on, she says: “These are dinosaurs that need to become extinct.”

Once a preanalytic automation system is installed, she predicts the lab will be able to recoup costs rapidly. “We’ll be able to streamline our whole area with fewer instruments, currently from 12 plating instruments down to three or four; there will be less maintenance to worry about; and the newer systems are more fully automated and can plate a wider range of specimens.”

They also believe that automation produces a better quality of product. “You reduce that human variation. For us automation is going to standardize plating, making sure we have the right inoculum each time.” The impact on turnaround time is not a primary consideration. “In theory, it could reduce turnaround time if specimens are being batched to plate, but the difference is probably marginal when you are talking about plates with specimens that are sitting 18 or 24 hours before you can really look at them anyway.”

An added benefit of automation that applies particularly to large labs is the ergonomic impact. “Eliminating someone removing a boric acid tube cap 1,500 or 1,700 times a day, which is our urine culture volume, should translate to fewer repetitive motion injuries,” she says. Biosafety is another element. “We plate everything under a biological safety cabinet, and this will allow us to get rid of some of those cabinets. We’ll be able to redo our whole space configuration because we’ll have fewer instruments and won’t need as many hoods.”

Much of the interest in automating specimen processing will depend on the volume of testing a microbiology lab is doing, Dr. Novak-Weekley says. But that’s precisely the reason that Steven D. Dallas, PhD, D(ABMM), assistant director of the microbiology laboratory, University Hospital in San Antonio, is more skeptical about automation’s future. “I don’t think it’s everybody’s dream to automate the front end,” he says, adding that University Hospital has no immediate plans to do so.

“Microbiologists are used to doing things manually, they’re very visually oriented, and a good one can work up and down the scale of tasks in the lab. You get this automated thing in the lab, it can only do that one thing. It can’t make reagents, answer the phone, or cover for me while I’m on break. So I haven’t heard too many microbiologists going, ‘Wow, I wish I could have a machine to replace this staff.’”

There are certainly advantages, he notes, to having a machine operate consistently. “These systems make really pretty streaking and you get better isolated colonies. Techs can do the same thing, but when techs are in a hurry, they get sloppy and put the ‘Mark of Zorro,’ where they just do this little zig-zag streak and the colonies aren’t well isolated, and then you have to go back and re-isolate. And that costs the patient another 24 hours of diagnostic time.”

The variation in different microbiology specimens will continue to be a stumbling block, in Dr. Dallas’ view. “Specimens that are really uniform like urine will do the best with the current systems. But there’s really not going to be a way in the future that I can see to consistently process every specimen that we have to culture. For many sample types you can have multiple different types of containers. Then if you have something like a liver biopsy for fungus culture, it’s a very, very small piece of tissue that the pathologist sees first; then when it goes to microbiology, it has to be manipulated. It’s usually not ground; it’s macerated into pieces and embedded in the agar. There’s no way an automated system is going to be able to pick up little liver pieces and embed them into fungal media tubes or plates.”

Although his laboratory is open 24 hours, it has about eight people during the day shift, one or two on the second shift, and one on the third. “So their functions go down to doing tests that are stat or have to be tended to right away. I think that might be a factor in how quickly automation could be adopted. Even if we had a specimen-processing instrument, if just one culture came in, such as a spinal fluid, it would be quicker just to plate it rather than fire up the instrument, punch in the keyboard, turn on the interface with the LIS, and so on.”

Dr. Dallas terms automated specimen processing an “80 percent solution.” “The analogy I use is the hotel housekeepers’ vacuum, with no attachments. It will vacuum really well in the middle of the room, but won’t do such a good job around the corners. That’s how a lot of these systems are going to be.”

He participates in chat groups for microbiology directors, and his impression is that front-end processing is not a hot topic. “They’re excited about automated PCR, about predicting resistance using molecular probes, and about spectrometry—the middle end, which is the ID part, and the back end, which is the susceptibility testing. Not about the front end.”

Microbiology labs in Europe have both advantages and problems that have made them earlier adopters of front-end and total automation. “Europe is more advanced on automation partly because it has a very unionized workforce,” Copan’s Sharples says. “In the U.S., you can get creative and hire someone who’s not necessarily a medical technologist, but not in Europe. So their labor crisis is even more acute than the U.S.’s.”

Stricter European requirements for screening of patients for HAI (health-care–associated infections) or MRSA also often apply. “They’re more demanding than in the U.S., and some labs have even made the argument for automation simply based on HAI and its effect on workload,” Sharples says. Because of European lab accreditation requirements, traceability of specimens is important, and that favors automation as well. “Plating and streaking has been a very manual task, and the risk of transposition and transcription errors is huge; manually done, they don’t have traceability.” With automated specimen processing like WASP, Sharples says, bar codes are scanned, labels are produced automatically, and date and time stamps show exactly when the specimen was handled and plated. “Everything is precisely traceable.”

If the adoption rates in Europe are a bellwether for the U.S., however, prospects for front-end automation sales in the U.S. are good. When his hospital in Bruges, Belgium, decided to automate a few years ago, says Bart Gordts, MD, who at that time was head of microbiology and infection control, few systems were available. “Before we decided to buy WASP, we had been looking for about six months.” That Belgian hospital started routine activity with WASP in November 2009, and Dr. Gordts is again researching automation for his current laboratory at Ziekenhuisnetwerk in Antwerp, where he is clinical microbiologist and coordinator of infection control.

“We were very happy with the WASP system,” he says. Like many other Western countries, Belgium suffers from a lack of skilled medical technologists, and that problem became especially serious when the hospital decided to change its infection control strategy by routinely screening all incoming patients for MRSA. “We calculated we would need to run an additional 40,000 to 50,000 samples a year, and due to the shortage and labor costs of skilled technologists in Belgium, “automation was the only way we could introduce that strategy of screening.”

His hospital was the first one in Belgium to conduct this screening for all patients, and it has had striking success. As his team reported at the recent Liverpool meeting of the Hospital Infection Control Society, “Our MRSA incidence of new infections dropped by 70 percent.”

Nurses are expected to ask a list of questions to determine whether patients have a risk factor for MRSA carriage, he says, but in a study his hospital found poor compliance: Only about half of the time were the questions being asked. That was an argument for systematic screening. “It’s much easier and more practical for hospitals to screen all incoming patients rather than having a nurse or doctor check for risk factors.”

Still, the laboratory had to face down controversy to take such a sweeping step. “Not many labs in Belgian hospitals are ready to process up to 50,000 more samples a year, and they were also afraid of the costs. In addition, there has been a lot of discussion among some microbiologists who would prefer to screen patients on admission with rapid PCR techniques rather than classic culture.” However, a multicenter trial in which his hospital participated demonstrated that PCR screening would not be cost-effective, he says.

The standardization of the inoculation pattern is one of Dr. Gordts’ favorite features of WASP. “Before we had WASP, the next day one could almost see which technician had inoculated which sample; there was such a big variation in how they do it.” If the techniques were good, “I would have been happy, but they weren’t always good.”

He has found the sterile loops WASP uses to be cost-effective. “The loops are reusable and allow us to inoculate two different samples on one plate. If you plan to have an additional 25,000 or 30,000 plates for MRSA, and you can plate two samples on one medium, that makes a difference of 50,000 euros a year.” He prefers loop technology to disposable combs, which he says are expensive to acquire and to get rid of, as they create medically dangerous waste. The beads technology that Kiestra is offering looks promising to Dr. Gordts, but he and other microbiologists in Antwerp are still gathering information about the pros and cons of all the methods.

“There is no way, in my opinion, that hospital microbiology labs cannot invest now in the automation of inoculation, because it really is the first step, and the evolution will be so quick,” Dr. Gordts says. He is convinced that in two or three years labs like his in Bruges or Antwerp will be entirely different. “I won’t claim we will operate without technologists, but we will be able to start with new techniques without investing in more lab technologists.”

Sharples is confident that microbiology labs both large and small can make the transition to front-end automation. Copan’s WASPs are being used now in labs with as few as hundreds of specimens a night up to labs with 3,000 specimens a night. “We‘ve been very, very careful not to simply make our machine for big reference labs or big medical centers. They have the volume; they need automation. But we wanted to appeal to the small general hospital as well as big core labs, so the WASP is an open platform, it’s designed to have a long lifetime in the lab, and its modular nature will make it adaptable to new technology.”

There will be more and more buy-in for the concept of total laboratory automation, though one of the biggest challenges will be access to and availability of capital, predicts Dr. Ledeboer, who is joining Dr. Bourbeau to lead a workshop this month at the American Society for Microbiology general meeting, “Microbial Specimen Processing: Yesterday, Today, and Tomorrow.” “We’re talking about a substantial investment on the part of the lab to do this, and convincing hospitals that there’s a return on that type of investment is going to be critical to the success of this type of implementation,” he says. But the downturn in the economy has added pressure on labs to consider automation. “If you need to do more work with fewer people, automation will offer you that opportunity without having to address additional human equity.”

Will front-end automation eventually become the norm in the microbiology lab? Dr. Bourbeau believes the answer is yes. “People are going to realize these systems are in the mainstream. That hasn’t happened yet, but we think there will be a snowball effect, and definitely in 10 years there will be many, many more of these instruments in the lab.”