Would you drive your car the way you run your QC?
Is our QC asleep at the wheel? QC and driving share many of the same characteristics - it's a task we do constantly, one that most of the time is safe and, frankly, can be boring. But in both cases, we need to maintain our vigilance against the danger of an accident or mistake.
Would you drive your car the way you run your QC?
Recently, I was reading Tom Vanderbilt’s book Traffic: Why we drive the way we do. This book examines a lot of the frustrations (congestion, late merging) and dangers (distraction, cell phones) of driving: “Driving, for most of us, is what psychologists call an ‘overlearned’ activity. It is something we’re so well practiced at that we’re able to do with without much conscious thought. That makes our life easier, and it is how we become good at things.” Unfortunately, if we become so good at driving, sometimes we “zone out” while we’re driving. Vanderbilt lists the various names associated with this problem: “highway hypnosis”, “time-gap experience”, “microsleeps” at the wheel. The terms all point to the same thing: the problem with becoming so good at driving is that our attention wanders. Driving becomes so automatic and reflexive that we can drive without seeing dangers as they approach. This problem is called 'inattentional blindness.':
“Inattentional blindness, it has been suggested, is behind an entire category of crashes in traffic, those known as ‘looked but did not see accidents.’ ….[D]rivers were looking directly at a scene but somehow missed a vital part – perhaps because they were looking for something else, or perhaps because something came along they were not looking for.”
Does any of this sound familiar? Do you think our laboratory quality efforts are another “overlearned” activity? After all, quality control is a process we do every day, just like driving. Most of the time, like driving, there aren’t any dangers, so it’s easy to ‘zone out’ and miss problems even as they occur right in front of us. Reading through Traffic, I began to wonder, what else are we doing in QC that is similar to these driving behaviors? And how crazy would it be if we transferred some of our QC behaviors over to the way we drive our cars?
Would you drive your car at the same speed, whether you’re backing out of your driveway, driving down a country gravel road, searching for a parking spot on a local street, or cruising on a highway? Would you drive your car at the same speed regardless of weather conditions like rain, snow, blizzard, monsoon, or plain old broad daylight?
Many laboratories adopt a “single speed” mentality when it comes to quality control. They treat every test method in the same way. Manual methods and highly automated 6th-generation methods get the same quality control approach. While this is mentally-efficient – we don’t have to spend any thought on considering QC – it’s easy to see that this is not resource-efficient. With QC design and Sigma-metrics, we can quantitatively assess and optimize QC procedures for laboratory tests. For tests with great performance and/or wide tolerances, we can reduce the amount of QC effort through fewer controls and wider control limits.
Years ago, the capability of most instruments and informatics for QC was basically “one-size-fits-all.” If you had an instrument with 50 tests, there was only one choice for QC and it had to apply to all the tests. Thankfully, that time is past. Many new instruments and software programs (be they middleware or LIS or something built into the instrument itself) offer the ability to customize the QC for each test. If you aren’t working in a lab that has that feature, you need to demand it from your vendors. The “single speed” QC approach is over-controlling some methods and under-controlling others. Neither of these approaches is good for the laboratory or the patient.
Would you assume that if your “check engine” light isn’t blinking, that means you’re a good driver? Or if your "door open" or "trunk open" light isn’t on, that also means you’re a good driver?
Today’s automobiles have a lot of built-in function checks. There are computers on-board that are monitoring hundreds, even thousands of operations in the car. Nevertheless, each function check has only a narrow perspective: one single part or process of the total driving process. None of those individual checks can substitute for an evaluation of the driver. They don’t license cars, they license drivers. So when you apply for your driver’s license, the evaluator doesn’t rate your car dashboard, he or she watches you.
Think of the total testing process in the laboratory. Our instruments have many function checks – this is often what we call “electronic QC.” But while those function checks are helpful, they are not sufficient. If a voltage check on a meter is working, that check is helpful. That doesn't mean everything is fine or that the current operator is competent to perform the test. "Traditional" QC is a measure of both the instrument and the operator together and gives us a better picture of whether the test can be performed correctly.
When you purchase a new car, would you let the dealer do the test drive for you?
One of the recent trends in laboratory purchases is the out-sourcing of method validation studies. That is, when a laboratory purchases an instrument, they require the manufacturer to run the method validation studies during the installation of the instrument. Usually, this is a service that gets "included" as part of the purchase price. On the one hand, this feels like a value-added activity - the laboratory gets something extra for no additional cost. But the service technicians have every motivation to get the instrument to "pass" its validation, and they might not perform all the studies that good laboratory practice dictates. Worse still, if the laboratory lacks the skills or manpower to perform the validation studies, does this also mean that the laboratory lacks the skills to interpret and understand the validation results? It's a chilling testament to the challenges facing the laboratory professions: we're losing the basic skills to be able to understand what we do.
Method Validation should be a core competency of the laboratory. It's understandable that laboratories might not have the staff to perform validation studies (there's no lab that isn't challenged for staffing and time), but each laboratory should still have the skills to dictate the scope of these studies and make critical judgments on the meaning of the results. Validation is not the process of generating statistics and charts to satisfy regulatory rules and stay in compliance. It's the process of making sure your method is working appropriately and applying your clinical judgment of the acceptability of the test.
Would you every consider reducing your seatbelt use to once a week or once a month?
Here’s a case where we really need to give thanks and credit to Dr. Jan Krouwer, who many years ago noted the absurdity of the EQC approach by making this analogy. If you think the EQC evaluation protocols were valid, he asked, why not apply them to your personal use of seatbelts and airbags: "If someone took out the airbags in their car and stopped wearing seat belts and didn’t get into an accident, they might claim that one doesn’t need airbags or seatbelts because they have had no injuries without their use."
Here's EQC thinking for seatbelt use. Option 1: if you drive for 10 days and find that you haven’t been in a car accident, reduce your seatbelt use to once per month. Option 2: drive your car for 30 days. If you haven’t been in a car accident, reduce your seatbelt use to once a week. Option 3: drive your car for 60 days; if you haven’t been in a car accident, reduce your seatbelt use to once a week. Oh, and if during the evaluation protocol, you find that you were in a crash, then you just need to re-start the evaluation protocol, not conclude that a reduction in seatbelt use could be dangerous
Obviously, it’s absurd on its face. If we only effectively attempt to detect errors once a month, what happens to those errors that occur when we aren’t trying to detect them? We lose the effectiveness of the safety procedure. We'd never accept the results of EQC protocols if our lives so directly depended on them. That's part of the reason why CMS was never able to scientifically justify EQC, and why EQC will eventually be eliminated.
As we consider the future Risk Analysis approach to determining QC frequency and effort, we can extend the analogy further. If you’re purchasing a new car, with new road-sensing technologies and smart-steering capabilities and other novel safety technologies, does that really mean you stop wearing your seatbelt and rip out the airbags? Clearly new technologies are reducing the risk of driving, but there is some basic level of safety that we still need in place, something that will always be in place to catch the error, whenever it might occur.
If your engine begins to sound like it’s having a problem, would you just turn up your radio?
This is similar to our “repeat the control” reaction to out-of-control events. We’re trying to ignore the problem by repeating the control, or running new controls, or repeating new controls enough times that we finally get one value to fall back “in” – so we can then ignore all the previous “out” flags. In effect, we’re pretending there’s nothing wrong with our system until the error grows to be such a big problem that it finally drowns out our attempts to ignore them.
I have to admit I’m guilty of this behavior. When I was just out of college, had my first car and my first (low-paying) job, I turned up my radio more than once. I didn’t have the budget for an expensive car repair at the time, so I was trying to ignore the problem, hoping it would go away, or at least postponing the day of reckoning. The slipping fan belt finally got so loud that it exceeded the volume of my stereo system. At that point, I suppose I could have bought a louder stereo, but I chose instead to finally have the fan belt fixed.
This kind of behavior is common in both car owners and lab operators. It’s not so bad when it’s only you and your car that are affected by that behavior. But usually the danger isn’t confined to our libertarian self – how we behave affects others on the road. When we’re a danger to fellow cars and drive, our behavior isn’t as acceptable and it begins to take the shape of negligence. When we ignore persistent problems with our QC, it’s an inconvenience for us, but it could potentially be a much bigger problem for our clinicians and patients. By the time we admit there’s a problem with our QC and pay attention to it, it may be too late. Those test results might already be out the door and in the hands of clinicians making diagnoses and therapy decisions.
Sometimes, a different perspective helps snap us out of our highway hypnosis. There’s a motto used by the "Click-and-Clack" brothers at the end of their radio show (Tom and Ray Magliozzi are the hosts of the NPR’s popular “Car Talk”): “Don’t drive like my brother.” In the laboratory’s case, perhaps we also need to invert it: Please, run your QC like you were doing a laboratory test for your brother.