AP labs diving deeper into automation
Posted on: Tuesday, September 24, 2019 By: KorchekStaff
September 2019—Talk to a few anatomic pathology laboratory directors about automation, and you may hear references to early network television, when automation’s downsides were mined for comedy. In one interview with CAP TODAY, a lab director drew parallels between potential backups in the automated AP lab and Lucy and Ethel’s travails keeping pace with a chocolate factory conveyor belt.
There is strong agreement, to state it in 1960s TV terms, that even for core or centralized AP labs with the necessary volume, the traditional automation options have progressed well beyond the modern Stone Age, but reaching Tomorrowland may require a shift in thinking.
For Jeffrey W. Prichard, DO, the image of the Jetsons’ flying car fits the futuristic automation of clinical pathology, which already has fully automated lines that handle specimen identification, preparation, testing, storage, and retrieval. In AP, says Dr. Prichard, division director of anatomic pathology operations, informatics, and quality for Geisinger Health System, Danville, Pa., only the most standardized and predictable processes have proved to be amenable to automation.
“But AP automation opportunities are there when you look for them,” he says. AP laboratories should consider not just full automation but partial automation as well, and they should look beyond mechanical robotic automation, where innovation has been slow, and consider data automation and the benefits it brings, Dr. Prichard says.
At his core laboratory, “We have automated our tissue processors and we have automated our slide stainers. We load the slides on and both the slide and reagents come off at a certain time. The stainers took over work that requires a few FTEs to complete, so I consider those fully automated, and I think that is analogous to trusting a fully automated car to drive up and down a six-lane highway like the I-95 to Florida,” Dr. Prichard says. But for the many variable or delicate processes in the AP lab that remain largely manual, automation is a trickier proposition. “I think the technology is not necessarily good for making it into my parking lot space.”
The progress he has witnessed in the past five or 10 years has been in the area of partial automation that still requires human interaction and oversight. He compares the complementary automation features arising in AP to driver’s assist functions in automobiles, such as cruise control, drowsiness and lane detection, and emergency braking.
His laboratory has consolidated six other labs at Geisinger over the past 10 years and runs couriers for the system of nine hospitals in Pennsylvania. One aim has been to reduce FTEs. “But we have not been able to realize that because our system has been growing so rapidly,” he says. In fact, he has not been able to recruit permanent histotechnologists fast enough to keep up with the volume, having to rely instead on temporary help, which is costly.
“We shoot for a turnaround of 80 percent in two days and in the past we have averaged 84 percent to 86 percent, but I have seen it steadily dropping to 80 percent this past year. I am attributing that drop to the slowdown in histology, because it is not that our pathologists haven’t kept up. The bottleneck has been in the manual processes in embedding and sectioning.”
The laboratory’s setup right now has mechanical automation starting with six Sakura Tissue-Tek VIP processors and two Leica Peloris processors. “We have been moving from our bank of reliable VIP processors we’ve had for many years and are acquiring the Peloris processors as we have been replacing the old ones,” Dr. Prichard says. At the embedding and cutting stages, the process becomes manual. But the staining tests are all done through a Roche Symphony HE stainer and, for IHC, Roche BenchMark Ultras and Leica Bonds. “Those I consider fully automated; they just need to be loaded and offloaded. You have to manually move the slides onto the coverslipper for the special stains in IHC, but the HE stainer has the integrated coverslipper.”
Because of the surge in volume at his lab, it hasn’t been difficult to get the budget to acquire new instruments, and he has not written off the technology that’s available for further automation. “I have been demoing these automation technologies at the trade shows. I just haven’t found one that actually saves me time,” even for the limited optimized type of control tissues used in demos. “I do look at the automated sectioners, and I am hoping they get more versatile and more efficient.” But right now, “It takes a long time to watch them cut a single section.”
“There is too much variability. Adapting to all the variant tissue, the density of the tissue, the amount of tissue in the block—that still remains part of the art of the histotechnologists. They can adjust the speed at which they’re sectioning in order not to compress the tissues and get good sections off of it. And that is not something that automation is yet able to achieve.”
Data automation has been the source of most of the labor-saving innovation in his laboratory in recent years, Dr. Prichard says. For example, “About eight years ago we fully automated barcode tracking throughout our Pennsylvania hospitals. There were still tissues being misidentified, and people were trying to read five-figure accession numbers off the blocks.” At one point there was a labeling error for every 5,000 blocks. But since switching to barcoding, “we’ve done 2.2 million tissue blocks without a block misidentification error recorded in the lab.”
Since barcoding is now 2D and contains much additional information, it has also automated workload distribution. “You need to assign cases to pathologists and ensure they have the right specialty training, you need to even the load among the pathologists with that training, and you need to know which space they are located at during the day. This was being done by accessioning staff in the lab. Now it is simplified, and it has kept our doctors happy that we aren’t overloading them compared to their colleagues.” (A sample warning: “Please note that on 09/09/2014 either the GI pod case limit for a pathologist has been reached or the daily case limit for a pathologist working in pod GI has been reached.”) Alerts to residents about specimens assigned to them go out by secure text to their phones. “That has been an efficiency for us and a crowd pleaser for the residents.”
An especially intriguing application is Dr. Prichard’s automating of quality management by creating structured data sets or multiple choice inputs for when quality errors occur. “Our process has multiple steps that are very manual and involve lots of people and lots of handoffs at multiple sites. Every handoff is a risk for the testing process,” he notes. To stem potential errors, he employed a kind of crowdsourcing. He asked all staff involved in the process and quality monitoring to provide input, when something has gone wrong at their station, through a structured data set that allows the system to count and trend the occurrence of quality errors. If an error has occurred, an alert graphic automatically displays on the monitor for that section of the lab. “Because of those, they have named me Dr. Redbar. I guess I am a little bit proud of that,” Dr. Prichard says.
The customized data input system allows the lab to report the 10 most frequent issues an individual has reported and provide feedback about the same errors occurring anywhere in the system. “I do that to try to engage the staff in the quality process, so that if an error happens to them, it shows how often it happens to anyone else and it includes whether management is doing anything to improve quality already.” There has to be a two-way street when crowdsourcing and engaging people in quality management, Dr. Prichard explains. Nine years of this system has produced 127,000 issues input by staff, he says, “which is an awful lot of crowdsourcing information we would not have had otherwise about the quality of our process.”
All of which has not hurt the bottom line. By automating collection of the data, “we have seen half a million dollars in labor savings by decreasing the incidence of these issues and not having to put labor into fixing them over time. Decreasing the amount of delays happening in cases year over year saved 31,000 days in delayed reporting of issues. Eighteen percent of those were inpatient cases. So, if even 10 percent of those delays happened with hospital and surgical patients for one day, that represents a $2.5 million savings in length-of-stay costs for the system.” That figure got Dr. Prichard a lot of attention from the C-suite, he says. “They are considering rolling this out beyond the lab into other areas of the hospital, such as the ICU, for managing their processes as well.”
Automation at AdventHealth Central Florida’s pathology laboratory, which serves nine hospitals in the Greater Orlando area, was installed about five years ago, but the lab started thinking about it several years earlier, says Debbie Dickman, MA, MT(ASCP)SH, senior manager of the pathology laboratory.
“We have always had some sort of automated H&E stainer, but everything else was very manual; we only started doing the special stains in histology and making sure we have good tracking from the computer side five years ago,” Dickman says. Now, the laboratory uses Sunquest CoPathPlus for its main tracking system and employs a variety of instruments: Leica Peloris for tissue processors, SlideMate slide printers from Thermo Fisher, a Sakura H&E stainer with an integrated coverslipper, special stainers from Agilent, and Leica Bond instruments for immunohistochemistry.
It’s true that labs must be a certain size to justify automation, says the laboratory’s histology manager, Jason Molnar, MHSA, HT(ASCP)CM QIHC. “You have to have the bandwidth, the volume, to be able to justify it because these automated instruments are quite expensive.” For his laboratory, the hopes of achieving economies of scale with centralization have been realized. The 35 technologists at the core lab process about 1,600 cassettes or roughly 350 cases a day. “If you put all the work back in the nine hospitals in our system and had to staff all those places, we would probably be closer to 50 people,” Molnar says. The lab can also handle steady volume increases—about 10 percent per year on average—without adding new staff.
The laboratory’s 2D barcode system allows tracking of specimens from point of collection in the procedure rooms all the way through the pathology process to the time that the specimen container is discarded and until the time the slides are discarded or sent to research 10 years later. “We track all steps in our process. In the laboratory where we process histology, we have 10 cutting stations, four embedding centers, and two distribution computer terminals to scan and track our blocks and slides,” Molnar explains. “So we have about 15 scanning apparatuses in the main laboratory.”
“We use codes that have RFID tags that are picked up by our WiFi system, so we know where the specimen is at all times,” Dickman adds.
But achieving a smooth workflow is challenging, Molnar says. “Realistically speaking, we are a full-service hospital, and it is difficult to get everything out because we handle all the makeups and specimens that other small reference labs or even larger reference labs that deal with biopsies don’t have to deal with.”
The lab considered acquiring a new Roche stainer this year but, for practical reasons, ended up staying with the relatively older stainers that are not interfaced with the LIS. “Once our slides come off the stainer, we triage them. Then we put them back together because as a full-service hospital we get larger specimens and they usually process them at different times and in multiple parts. Then we verify in the system that we have all the parts for that case before we distribute it to our pathologists,” Molnar explains.
They would like to see a continuous workflow, says Dickman, but are working with sites that are one hour away, waiting for specimens to come out of the procedure room. “The majority of our work comes in the afternoons, so we have these big pulses of work we are trying to push out.” The manual steps are many, starting with grossing. “Then the tissues come off the processors at different times.”
“I know there is more automation available out there, but I don’t know how much is practical for a high-volume lab like ours that has a large variety of case mixes,” she says.
One of automation’s gifts is the ability to better monitor laboratory performance. “For routine specimens, we like to have 90 percent of our cases signed out in 48 hours, and we have a lot of cases signed out in 24 hours if they don’t require additional testing,” Dickman says. “Automation has allowed us to capture that turnaround time information in the grossing room or the time from grossing to embedding. It helps us locate slides that were misfiled, and it helps us find those blind spots where we can improve or where we might have waiting time that we will continue to work on.”
The lab is in the reviewing phase of acquiring Thermo Fisher’s Arcos system to automate the block archival process. “Currently we have to spend our manpower to manually file blocks and slides when they are returned to us or when we are finished with them,” Molnar says. “But because we have nine different locations in our system and each one has a different accession prefix, they all have to be filed not only in order by number but also by location. And that is something we do manually.” Sometimes it can take up to an hour to find where a block is. “We’re hoping to onboard the new system by the end of the year and hoping it will improve our processes.”
To maintain productivity, the lab has to simulate a continuous flow and does that by batching cassettes. But Molnar says they are evaluating batch size because it can create bottlenecks at different locations in the lab. “The situation in I Love Lucy where the candy on the conveyor belt is going all over the place is what we want to avoid at all costs. We tell staff—especially the third-shift staff because that is when the majority of our work hits the laboratory—they have to be aware they can’t just push and push from embedding to microtomy because the work starts to back up. We have to be vigilant to avoid that.”
His laboratory is large enough to move further into automation, he says, and the next step might be embedding automation, but only if the technology can be shown to produce to the same level of a histotechnologist and have the same professional judgment. “That is where you have to ask yourself: How far are we going to go with automation in the AP laboratory?”
Dickman advises openness to automation, but also caution: “I think you need to be open to the potential of automation and look at the barriers that vendors have and see if it is something that will work for your laboratory.”
Denise Bland, MHA, CTBS(AATB)HTL(ASCP)QIHC, senior technical director of histopathology at Massachusetts General Hospital, has been engaged in automating her lab for the past 10 years, but growing volume has not been the only driver. The MGH lab performs about 800,000 tests a year, on specimens from all over. Volume has risen about 10 percent since 2014.
“The need for automation isn’t only volume driven; it is also cost-of-medicine driven. It is value-based care,” Bland says.
Histotechnologists remain hard to find, and her staffing level is about the same as five years ago despite rising volume. “Automation allows us to not increase our staffing when we are able to get the automation we are hoping for and need. When the automation updates and advances progress, across the board we are able to maintain and let our technicians focus on more quality tasks than what the automation should be doing.”
Full-line automation could eventually become a reality, Bland believes. “The workflow is barcode driven and the vast majority of labs today have barcoding. From the fixation station, the specimen can pass to a processor that passes to an embedding center that passes to a microtomy station. And from the microtomy station, the system would know from the barcode and LIS protocol dictionary which staining system it needs to go to, or it will pass it to the archive system if it is unstained. You can see where it could all come together. But no one vendor is bridging the gap yet. They have to have more motivation to get there in a cost-efficient way.”
Theoretically, the electronic health record and LIS could drive this imagined full-line automation, as long as vendors will allow open interfaces, she says. “But there also have to be transfer stations that are either from the same vendor or that connect to other vendors’ stations. So there is also vendor cooperation that needs to be going on behind the scenes for the benefit of the workflow. Think of standardized accessories across platforms, such as processor racks or staining racks.”
Unfortunately, Bland says, the automated microtomes on the market have made little progress in getting to the point where they can replace histotechnologists. “I know units are being used in the research world, but the clinical world is where we need the turnaround time and efficiency for value-based medicine and care.”
She has hopes that future labs will become more artificial intelligence and deep learning focused, with digital imaging driving much of the workflow. “But that will not right now somehow replace slide generation. What happens after the slide is generated can be up for discussion, but we will still need to process, embed, and cut, in my mind.”
The lab maintains strict TAT guidelines using the Leica Bond IHC stainers and platforms for special stains, including Pathcom and the Leica Spectra and Roche BenchMark. “The special stain platforms are still significantly lacking improvements,” Bland says, “as much of the vendor focus is on more advanced staining, despite necessary volume in this area. Perhaps it will be digital imaging that forces innovation.”
In addition to the Leica APiQ reagent inventory control application, the most recent automation upgrade her lab has made is archival management through installation of Thermo Fisher’s Arcos, which has been a huge asset, she says. “Basically, we no longer have to put our blocks in order or keep them in order.” While the lab does organize blocks by year, “we put them in a tray that is scanned, and we can use a personal digital assistant to find and pull out blocks as we need to.” The PDA is like a FedEx PDA, she explains. “I’ll say, ‘I’m looking for block 1234’ and it will say, ‘OK, cabinet Z, 88.’ No one has to manage those assets the way they historically did. Not having somebody stand around all day to put blocks in order is huge, and we are about to roll the same thing out for our slides.”
This also allows for a high-security mechanism in that only those with access know the location of the asset. “This vendor also has the ability to know our error rate near real time via open interface and continuous communication. Perhaps this should be information all vendors seek and all lab managers should make available to them. Quality improvement through transparency and data sharing,” Bland says.
The anatomic pathology lab at Baylor Scott & White Medical Center, Temple, Tex., the core lab for a third of Baylor Scott & White Health’s 50-hospital system, is experiencing rapid growth as it takes on work from across the state. So far that includes 14 hospitals from Waco to south Austin. Last year, the lab had 70,000 cases; this year it is on track to reach 80,000 cases, including 300,000 H&E slides. Over the past five years, “I would say we have increased our volume 10 to 15 percent and added maybe one or two people to what we originally had, says Ary Franklin, HT(ASCP), CSSYB, histology and immunohistochemistry lab supervisor. “To increase volume that much with a minimal staffing increase has been a huge thing.”
The automation process started about 10 years ago with two automated stainers for H&Es, three automated stainers for special stains, and 2D barcoding. Today, “barcoding is in place in the gross lab, in histology, and IHC, and extends into molecular. While it cannot be used on the Agilent instruments, it is in use on all the lab’s Roche Ventana instruments.” That not only sharpens patient identification but also helps trace the source of adverse events more efficiently. “We look at a number of recuts and are able to identify which tech cut them so we can do retraining where necessary or maybe have a staff member realize they need to slow down,” Franklin says. Turnaround time has decreased to 1.5 days or less on the smaller cases—a sharp drop from a nearly 2.5-day TAT before—and there is a TAT of four days on complex cases.
The instruments now include two H&E stainers (Roche Ventana HE 600), three Roche Ventana BenchMark special stainers, one Agilent special stainer, two Agilent IHC stainers, one Roche Ventana IHC Ultra stainer, and a Leica IHC stainer. Ari Rao, MD, PhD, pathology chair, says each stainer has its unique strengths. “In IHC, we specifically have different vendors because there are different antibodies that work better on different platforms, or it may be because of the FDA approval process or because one works better in some situations and only one vendor is able to provide that. Leica, for example, works really well for us because it can be used for the longer runs. It is pretty hardy that way.”
Centralization of AP lab functions into one lab, though it entailed the expense of automation, was never in doubt. “It was much more cost-effective to add the current work to us at Temple and have it centralized than to add a histology lab to each of the other facilities. A lot of our work has come from new hospitals that were built at different locations,” Dr. Rao says. The system conducted a cost analysis that showed a savings of $10 to $20 per block if specimens were sent to the central hub.
One of the bottlenecks the lab encounters is with the stainers, Franklin says. “I love the Ventana HE 600s. They’re super-fast, and for the volume we put through them they do a great job. But because it is all a closed system, if something goes wrong, the technology is so integrated that you have to call in the company. And while they are timely in their response, it still causes a delay. With the typical dip-and-dunk stainers, you can manage it manually if you have to.”
An automated embedding station such as Sakura offers is on Franklin’s wish list. “For larger surgical tissue, that would significantly free up some of our techs to do microtomy instead of embedding,” she says. She has seen an automated microtome where a machine puts sections on slides but worries that the technology is still too new. “We have a semiautomated one that can trim your blocks, which is great. But if you are cutting really minute tissue and having to decide how much tissue do I have left and what do I need to do at this point,” she is not sure that can be managed with automation.
The paper steps that remain in the AP lab are at the start and end of the process, Franklin says. “For the most part, all the lab orders have a paper requisition that comes with the specimen. And the only other time paper is involved is at case assembly when they are assembling the slides” and designating where they are to be sent. When digital pathology comes online—the lab’s next largest upcoming investment—she expects the slides will be scanned automatically and the paper component will be phased out.
Dr. Rao thinks automation has delivered as promised, for the most part. “It has taken a little longer for some of the processes.” For example, “Getting the stains just the way we like them took a little longer than we thought. Then we had to work to optimize the stains so that the pathologists would feel comfortable with them. After that, the next setup of instruments was much easier.”
Franklin believes the distance that some of the histotechnologists are willing to commute to Temple—up to 45 minutes to an hour away—is an indicator of how much they like the centralized, automated lab. “I think it is a draw for those histotechs who have experienced it. They enjoy the automation and innovative technology.”
To decide whether to centralize and automate, the average AP lab needs to weigh the alternatives, Dr. Rao says. “First, you really have to have a good idea of what your expense would be if you didn’t centralize,” she says. “We projected a cost for each of these facilities to have its own lab, then showed the cost of adding just a little bit of capacity at the central lab, and that savings was quite dramatic. The second piece, I would say, is charting out the entire pathway to automation.” That is doable, in her view, because automation in histology is more mature than it used to be.
Once these factors are taken into account, Franklin advises AP labs that are on the fence about automation to take the plunge. “You should absolutely go for it. I don’t remember a single loss of a specimen since we had automation. It improves your quality, and the efficiency of getting results in a timely fashion is amazing.”