Tools, Technologies and Training for Healthcare Laboratories

Pre-, Post- & Analytical Errors

Which improvements should be made first? Unfortunately, we seem to be arguing that some errors are more important (i.e. worse) than others. But rather than make a Chicken-or-Egg choice, David Plaut, Dr. Westgard, and Sten Westgard urge an analysis of these types of errors. The answer to the question of which improvements come first need not be "pre" or "post" or "analytical" - it should be "all three at once."

Making sense of conflicting priorities in the laboratory

We've often heard the opinion that the Quality Control of laboratory testing isn't the biggest problem we're facing. Sometime people quote a statistic that 40% of the errors in the laboratory are pre-analytical, 40% are post-analytical, and "only" 20% are analytical. There are more "P-errors" than "A-errors", therefore, many laboratories believe they should put a higher priority on pre- and post-analytical errors than on analytical errors.

First of all, you may find it depressing when we admit we've got problems everywhere and we only argue over which problems are worse, which problems will be fixed next, and which problems will be ignored for the time being. Ignoring problems is not a good thing. We should be incensed that we've got so many errors in so many different areas of laboratory testing and patient care.

The source of this commonly accepted knowledge about laboratory errors is an abstract, not a peer-reviewed paper [see the complete abstract in an earlier discussion on this website]. It seems that people want to believe that analytical errors aren't as frequent as the pre-analytical and post-analytical errors. This belief is part of today's quality compliance mentality that assumes analytical quality is okay if laboratories follow CLIA rules and regulations. Laboratories assume they don't have any problems with analytical quality. If forced to admit they do, then they respond that there are other bigger problems that are more important, therefore they don't need to deal with analytical quality.

This "Sources of Errors" assumption threatens to become an excuse to avoid Quality Control. If we refuse to confront our analytical errors, if we postpone improving them while we work on the "bigger" problems, we let the QC problem fester and grow. The longer we wait, the worse the problem gets.

No one denies that there are pre- and post-analytical errors in healthcare. Nor can we, unfortunately, say that these errors are small or insignificant. However, we must challenge the notion that analytical errors are the "smallest" and therefore least important problems we face in the laboratory. We hope, by examination of the problem and by a few crude analogies, we can convince you of this.

Whose Errors are More "Obvious"?

What's the most obvious error?
1. An patient sample that doesn't get to the lab.
2. An instrument with a systematic bias.
3. A test result that gets reported on the wrong patient.

Pre-analytical and Post-analytical errors tend to fall into the "obvious" categories. When patient samples are improperly collected, mislabeled, aren't delivered in time, or are lost, it is obvious that some sort of error has occurred in the pre-analytical stage. When the results of a test aren't reported in time, or if they're entered into the wrong patient record, it's also pretty obvious.

Perhaps we should unify the pre-analytical and post-analytical error categories. They fall into one bigger, more important category. It's the "This Error Makes the Doctors Angry" category. The doctor can quickly identify when a pre- or post-analytical error has occurred, and find fault with those "responsible" for the error. Since doctors throw a lot of weight around a hospital, their complaints become the most important. This is probably why pre- and post-analytical are believed to be the "bigger" problems. The doctors' shouting tends to be louder, and heard more, than any other voices.

In contrast, analytical errors fall into the "This Error CAN'T readily be detected by the Doctor" category. When the doctor receives the test results from the laboratory, all he or she gets is the numbers. There is no way to know if those numbers are biased up or down because of method or instrument problems, or if there is some strange random fluctuations that are throwing off the result. Perhaps the doctor might suspect an analytical error, if the test results are extremely divergent from all the other symptoms the patient is showing, but even then the doctor can only guess (and most likely, order more testing!). For analytical quality, the doctors are completely dependent on the laboratory for the detection and correction of errors.

It's common sense that patient samples should be collected in the proper way, and that they should be labeled correctly to clearly identify which patient they were taken from. It's common sense that patient results should be delivered to the correct patient and doctor, too. The reason these don't happen isn't a lack of understanding, it's an overload of information and traffic. Hospitals are strained by the amount of patient information that they must manage. Many times, the laboratory can't correct those problems - it's a hospital-wide systemic problem. Those kinds of problems likely occur throughout the healthcare process, including the examination and operating rooms.

A Half-Baked Analogy

In a crude sense, a laboratory works just like pizza delivery business. The pizza shack business can be broken down into three core processes:
1. Getting the Order
2. Making the Pizza Order
3. Delivering the Ordered Pizza

Now let's say that these three processes are handled by three different employees
1. Getting the Order - the phone jockey
2. Making the Pizza - the cook
3. Delivering the Pizza - the delivery person

What kind of errors can occur in this situation?
1. Getting the Order - an improper order (different ingredients from those requested), an improper address
2. Making the Pizza - spoiled ingredients, faulty oven, improper baking time, improper handling of food (dirty hands), etc.
3. Delivering the Pizza - wrong address, box goes soggy, can't make change, late delivery

I know that this analogy breaks down if you scrutinize it too much. Laboratory tests are not pizzas. But I hope this little illustration gives you a clearer view of post-, pre-, and analytical errors.

Which Errors are Worse?

Another question raised by the coexistence of pre-, post-, and analytical errors is this: which ones affect the patients the most? We think this is a triple dead heat. If you can't get the patient specimen to the lab, if you can't perform the test correctly, and if you can't deliver the results back to the patient, the consequence is the same: poor patient care. No error is worse than the other. They are all equally terrible.

Let's put it this way: if you solve all the pre- and post-analytical errors, but haven't solved the analytical errors, you are still delivering bad patient care. To be sure, the patient sample got to the lab correctly and the result came back to the right patient, but the number is wrong - it's biased or totally off. How can the doctor make a good decision with bad numbers? In many cases, the doctor is forced to run the tests against, or rely on his or her "judgment."

Let's apply this idea to the pizza shack: It doesn't matter if you get a pizza at the right time with the wrong ingredients, or a pizza with the right ingredients a day after you ordered it, or if you get the pizza you ordered with the right ingredients at the right time but it's cold. In all three cases, it's not what you wanted.

In the end, all the errors are equally bad. No error is worse than another. In every case, you don't get the pizza you want, and you don't get the medical care you deserve.

Extending the Analogy: Baking the Pizza - Which Improvements Come First?

How do you become the best pizza shack in town? A simple question. You make sure you answer the phone promptly and courteously, get the order right, make the pizza that was ordered and deliver it hot to the right place with change as needed. Each of these three parts is important. If you have a problem with one, it doesn't matter how good you are at doing the other two parts. You won't be the number one pizza shack - and you might not be in business at all.

In a similar vein, progress in the laboratory testing area needs to happen all at once. We must reject the notion that progress can only be come in one area at a time. The problems are too great to go slowly. We must make efforts on all fronts. Even if this means making small improvements in each area, a unified improvement effort will achieve better test results and better patient care than narrow efforts in either the pre-, post- or analytical area.

If you improve in just one area, you're still delivering bad results and bad patient care. If your pre-analytical errors are greatly reduced but you've made no improvement elsewhere, nothing is that much better. All you're doing is delivering bad test results faster.

Who's job is it to bake the pizza? The cook. Who's job is it to provide correct test results? That's us - the professional laboratorian. Just like the cook, we're the critical element that remains behind-the-scene. The cook isn't the one who delivers your pizza; you'll never see him or her, but you assume that they're washing their hands, keeping the oven at the right temperature, etc. Only in a few gourmet restaurants, on rare occasions, does the cook emerge from the kitchen to receive the thanks of the patrons for all the fine food. When has that happened in a hospital? No patient asks to see the MT who ran the test. Unfortunately, we're the invisible but irreplaceable workforce in the hospital. And our invisibility often makes our problems invisible to upper management. But just because others don't see us, doesn't mean we don't have the responsibility to make our concerns heard and seen.

A Final Thought: What's the Core Job of a Laboratory?

This is another simple question, but laboratorians give a wide variety of answers. The laboratory produces test results. You feed in samples, it feeds out numbers. The core job of a laboratory is to produce the correct test result. If we can't get the test results right, then we aren't doing our core job.

All three types of errors need to be addressed NOW. Each has different root causes and each may require a different approach to solve. In some hospitals the laboratory isn't responsible for drawing the sample and getting it to the laboratory. Thus the pre-analytical error isn't something that the laboratory can address alone. Similarly many laboratories report data through a computer and it may take the Information Services folks to help with post-analytical errors.

But if we in the laboratory aren't producing accurate test results, why are we there in the hospital? A random number generator would be more efficient and certainly cheaper. It's our profession to know all the details of testing and instrumentation and quality control. It's our profession to assure that test results are correct.

So let's do it. In the laboratory. With the floor staff. With the phlebotomists. With the IS team.

Do you want anchovies on that?