There was a time when doing a lab test was essentially the same thing as understanding how it worked. If you measured something, you probably knew when to pour in the reagent, what color was supposed to change, why it might have gone wrong, and what it should look like before a number went onto a screen.
Take osmotic fragility as an example. You basically just put red cells into progressively more dilute saline solutions until eventually the red cells burst. That has to be strangely satisfying, like popping every single bubble in a long roll of your brother’s bubble wrap.
I'm not suggesting this was necessarily a golden age of lab testing. Obviously, it involved a great deal of pipetting and timing, washing, and even staring, willing the specimen to do what you wanted it to do. But the method was difficult precisely because it was easy to understand. It was sitting right in front of you.
The modern laboratory is more efficient in nearly every measurable way. Reproducibility, scalability, and automation have made laboratory testing a model for the modern hospital. Contemporary laboratories can process absurd numbers of specimens, which is wonderful. But as the analyzers improve, fewer and fewer people are able to understand all the methods within them.
Automation is one of the problems, but another is that the modern laboratory is doing many more kinds of things than its historical counterpart. A single lab may contain ion-selective electrodes, immunoassays, nephelometry, clot-based assays, mass spectrometry, flow cytometry, chromogenic assays, molecular amplification and sequencing, and digital imaging – not to mention all the software, middleware, and algorithms to interpret all of those results. Even within one specialization, two tests might be sitting next to each other on the same bench but depend on very different analytical principles.
You need to be a biochemist and an astrophysicist, not to mention a computer scientist and materials specialist, to fully understand all of these different methodologies. Honestly, at some point, you start to wonder whether we should just hire Bill Nye the Science Guy or Professor Dumbledore to explain all the magic in the laboratory. So the problem isn't that laboratorians understand less. The real problem is there's just more to know, and it's not realistic to imagine that one person can know it all.
Expertise has been distributed, which raises two important questions. Do enough people inside the laboratory understand each method well enough to recognize when something is wrong? And is there anyone left who can go all the way down when the usual troubleshooting script stops working?
The weird thing is that the loss of this knowledge about methodology can happen without any sort of outward-facing performance degradation. In fact, the laboratory can be performing as well as or better than ever. Turnaround times can go down, error rates can fall, automation becomes more reliable, and the number of specimens processed per FTE goes up. The system can become more “successful” at exactly the moment that fewer people understand each individual method well.
That’s what makes this problem different from ordinary training problems. There isn’t an obvious failure telling us what knowledge has been lost. If an assay stops working, we understand what the problem is. But if nobody knows how to validate a method, troubleshoot an unusual interference, explain why two analyzers disagree, or recognize that a result may be technically acceptable but make no biological sense, this knowledge gap may sit dormant for years.
The laboratory, therefore, has a peculiar vulnerability: expertise can disappear before performance does.
That creates a problem for how laboratories are staffed and administered. Expertise is difficult to defend when its value appears rarely, if ever. An administrator can point to turnaround time, cost per test, FTE productivity, error rates, instrument uptime, or volume. Those numbers are real, and laboratories should care about them. But it is much harder to put a number on the value of having somebody in the room who actually understands why an assay is behaving strangely.
The cost of that expertise is visible every year; its value may only become obvious when something unusual goes wrong. Even the salary structure may be moving in the same direction. In ASCP wage surveys, the pay premium for an MLS supervisor over a staff MLS fell from about 21 percent in 2013 to about 16 percent in 2023.
That person can look inefficient right up until the day they are indispensable.
If expertise can erode earlier than performance, then the next question is what kind of expertise should be maintained, how much, and where should it live? It’s not realistic to expect every laboratorian to understand every assay at the same time and to the same depth. The modern laboratory is simply too complex for that. But it’s equally unrealistic to assume that vendor specialists or one senior tech for an entire system can cover everything.
The laboratory needs layers of expertise. Many people need enough knowledge and understanding to recognize when something has moved outside the ordinary. Some people need enough depth to investigate it, and somewhere in the system there has to be at least someone who can go much further down when the usual troubleshooting scripts stop working. That raises a staffing question: how much redundancy is enough? Is one person enough? What happens when they go on a cruise to the Bahamas for their 25th wedding anniversary?
And being practical on this point is critical because it matters for training, competency, succession planning, and method ownership. Staffing justification becomes especially important when budgets are scrutinized. Deep expertise is inefficient on a financial spreadsheet, particularly when everything is working just fine the way that it is. But when our last best person walks out the door because she’s eligible for retirement, we need to have the proper language to justify finding someone who can replace her.
And as our workforce ages, these problems are going to persist and even grow. The 2024 ASCP Vacancy Survey found that nearly 28 percent of core laboratory supervisors are expected to retire within the next five years. So, laboratories will need to become much more deliberate about how they preserve and protect their expertise. Historically, a lot of technical knowledge was acquired through repetition and work. People learned by doing, seeing failures, and troubleshooting. But automation has removed a great deal of that repetitive work, which is largely good, while also removing some of the accidental education that came with it.
That means laboratories need to be creative and start to learn on purpose. Not by forcing people to perform obsolete methods that have been automated, but by giving them structured exposure to the kinds of problems that routine work no longer provides. That could mean troubleshooting exercises, interference investigations, validation work even when it is not strictly necessary, or reconciliation between two methods when they don’t agree rather than simply accepting the discrepancy.
The goal should be to make the workforce much more resilient. To create a team with the breadth and depth to handle nearly every kind of error. The field should begin to catalog the kinds of failure modes that are likely to be encountered and expose individual labs to them to maintain competency.
In short, the modern laboratory may need to train less for routine execution and more for exception handling.
The laboratory needs to begin to understand less by hand and by luck, and more by design. We shouldn't preserve obsolete work simply because it once produced expertise, but we do need to preserve the knowledge that work created. Laboratories already plan for instrument failure, downtime, backup methods, power, supplies, and staffing. Expertise should probably be treated the same way.
We are no longer expecting every laboratorian to know how every test works. That world has probably gone by for good. The question is whether the laboratory as a system still knows enough: whether the right expertise exists somewhere, whether someone else can access it, and whether the next generation has been exposed to the kinds of failures that routine automation increasingly hides. Automation and analyzer complexity have made the laboratory much better. Accuracy has improved, turnaround time has gone down, and throughput has skyrocketed. The challenge now is making sure the laboratory does not become so good at routine work that it forgets how to handle the weird stuff.
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
