Closing gaps in preanalytics: How technology is driving smarter solutions

A lot can go awry along the pathway from a test order to analysis, and it’s long been known that most errors in testing can be traced back to a preanalytical problem. Undetected, these issues cost laboratories time, money, and — most importantly — compromise the quality of patient care.

Like many of the problems clinical laboratories face, a lot of hope is placed on better technology to vanquish one of the most painful weak points in laboratory medicine. Fortunately, both academic and industry research scientists are showing progress that promises to offer relief.

“We’ve made good progress in the last decade in safety and cutting down on preanalytical errors,” said Mario Plebani, MD, FRCPA, FADLM, professor of clinical biochemistry and clinical molecular biology at the University of Padova School of Medicine in Italy, and chief of the department of laboratory medicine at the University-Hospital of Padova, where he was also the dean of the medical school. “But we still have more we can do, especially when it comes to incorporating automation, new technology, and artificial intelligence.”

Detecting and avoiding stubborn errors

For most clinical laboratories, electronic solutions that replace hand writing and affixing labels have largely taken care of many kinds of mundane errors, according to Plebani. “We are much better now at accurately identifying patients and properly labeling tubes,” he said.

One of the more pressing problems that still plagues clinical laboratories, though, has been hemolysis. This is the case across different countries with different automation systems, he said, especially when analyzing potassium and lactate dehydrogenase in serum plasma. “This interference is problematic because elevated levels of these markers can lead to diagnostic errors.”

It’s a problem companies like Werfen have been working to tackle, said Annie Winkler, MD, the company’s chief medical officer. In 2023, Werfen gained FDA regulatory clearance for the GEM Premier 7000 with iQM3, a new blood gas analyzer that integrates whole blood hemolysis detection into the device. It enables clinical laboratories to assess whole blood samples for hemolysis “without adding sample value or time,” she said. For patients, that means not having to get a sample recollected and not having to wait in the emergency department for additional time for another potassium result.

The technological key was Werfen’s ability to integrate the separation of whole blood with optical sensors that detect the degree of hemolysis and flag problematic results.

Werfen has also been looking at how their digital solutions can bring value not only to preanalytics but post-analytical support, which Winkler calls lab or clinical decision support depending on the intended user. Technology could interpret data to uncover findings not possible today by looking at data on a clot curve, for example. “We recognize that there are more and more data getting added to clinical care, and we have the opportunity to potentially provide some digital solutions.”

Human errors in focus

An overworked and understaffed workforce can also lead to mistakes, an alarm bell that the Association for Diagnostics & Laboratory Medicine (ADLM, formerly AACC) has been ringing for some time.

Issues with preanalytical errors aren’t new, but they started to come into sharper focus as a “byproduct of all the staffing challenges” that laboratories have been experiencing, said Paavana Sainath, head of core lab solutions R&D engineering, diagnostics at Siemens Healthineers. A recent survey conducted by Siemens Healthineers and The Harris Poll, which surveyed 408 U.S. laboratory professionals in the healthcare industry in June, found that 57% of laboratory professionals were performing sample collection, regardless of their role. “We know from our conversations with customers that this isn’t necessarily a choice,” she said.

The report found that 39% of laboratory professionals rank limited staff to support laboratory operations as their greatest challenge, and that 5% percent of laboratory professionals reported that their lab had closed temporarily because of understaffing.

Most concerning, the study found that 22% of laboratory professionals reported personally having made a lower-risk error due to feeling overworked or burned out; and that 14% have made a high risk error due to the same circumstances.

Automation and the introduction of workstations can help by reducing sample management tasks, Plebani said. It can eliminate mundane preanalytic duties, freeing staff to focus on other responsibilities or simply easing their workload to prevent overwork.

Sainath emphasized how Siemens Healthineers has enabled its Atellica Solution (high volume) and Atellica CI Analyzer (low to mid-volume) machines with its Atellica Integrated Automation functionality, which has been able to consolidate more than 25 manual tasks in preanalytics, analytics, and post analytics into a single analyzer. “The capabilities you typically find in these high volume or mega labs are now accessible to the lab around the corner,” Sainath said.

Getting samples to their destinations faster

Another possible source of preanalytical errors is in transportation. Pneumatic tubes, which are common in large healthcare facilities to get samples from the bedside to the laboratory, are another opportunity for things to go awry, as is walking samples from point A to point B.

Samples can be lost or sent to the wrong location, answers sent back to the wrong department, which can mean errors in care, and lengthy patient treatment delays, said Beau Fulbright, national sales and service manager at the laboratory automation division for Sarstedt.

To tackle these challenges, Sarstedt developed the Tempus600, a direct transport system that doesn’t require packaging or bagging. It also can send samples directly into a lab automation system, so samples are processed immediately. The system is compatible with the major IVD automation lines, for example Roche and Abbott.

This will also lead to better patient care and lower costs, he added. “If you cut down the time that it takes to get samples to the lab, hospitals can make better decisions on treatment,” he said. “Of course, we all know how expensive emergency room visits are, so even from a financial standpoint and not just patient care standpoint, if you free up that emergency bed 30 minutes faster, that’s somebody else that can get into that bed faster.”

Can artificial intelligence make a difference?

Artificial intelligence (AI) might be seen as an angel or devil in the room, depending on which person is looking at it on a given day. But its potential for detecting preanalytic error is becoming clear, experts said.

The Siemens Healthineers-Harris Poll report found that 52% of lab professionals strongly or somewhat agree that laboratory automation is a threat to their job — but at the same time, 95% said that the adoption of automated technologies will help them improve patient care.

“They have that tension, but also feel they could do their job better if they actually adopted the technology,” Sainath said.

Plebani is enthusiastic about the potential of AI, and he believes it can revolutionize the detection of preanalytical errors. Plebani pointed to papers published in Clinical Chemistry (doi: 10.1093/clinchem/hvad100), International Journal of Laboratory Hematology (10.1111/ijlh.13820), and American Journal of Clinical Pathology (doi: 10.1093/ajcp/aqy085) showing how machine learning was able to detect interferences.

Another recent study in Clinical Chemistry showed that machine learning can help accurately detect intravenous fluid contamination in blood samples even without exhaustive pre-labeled data from experts to train the algorithm (doi: 10.1093/clinchem/hvad207).

As always, further research and data will need to validate and refine such protocols — but the benefit to patients adds urgency to the effort. “The auto-verification of results thought to be erroneous represents a stark departure from the current paradigm and would require extensive observation and validation prior to deployment,” the authors wrote. “However, with the recent advances in machine learning development and deployment and [laboratory information systems], combined with the nearly universal staffing shortages for laboratories and the fallibility of human adjudicators, the potential benefits to clinical operations and patient safety indicate the need for further investigation.”

Jen A. Miller is a freelance journalist who lives in Audubon, New Jersey. +X: @byJenAMiller.