Questions

By theory, after you get an out-of-control event, all of those patient specimens that were tested during the run must be repeated. But labs don't really do that.

 Reality Check on Patient Re-testing

December 2023
Sten Westgard, MS

An articulate set of questions came in to us recently, concerning the proper re-testing of patient specimens after an out-of-control run.

The textbooks and the theory of QC states that if the run is out-of-control, all the patient samples should be retested. In the dawn of the early age, this wasn't so difficult, because the patients were batched and often bracketed. So a run wasn't released or reported until those QC results were "in." Further, and this may be the bias of modern experience talking, the volumes in decades past were low enough that it was more possible to repeat all the patient samples.

The reality of today is that laboratories operate in continuous release mode, so patient results go out the door before QC is run. More difficult still is that the volume of testing is much, much higher today. So much so, there may not be any practical hope of being able to retest an entire run (most likely an entire day) of patients. There is no slack in the system to absorb a failure.

So the laboratory is in a quandry. How do they rationalize repeating fewer patients from an out-of-control run?

What the regulations require

ISO 15189 has a concise statement on the topic:

"7.3.7.s Internal Quality Control (IQC) ....
g)...1) When IQC defined acceptability criteria are not fulfilled and indicate results are likely to contain clinically significant errors, the results shall be rejected and relevant patient samples re-examined after the error has been corrected...."

Note the artful term "relevant" patient samples. So implicitly, this doesn't have to mean all patient samples need to be re-examined.

CLIA's SOM has a D-tag that doesn't directly adress retesting, but instead covers the broader concept of corrective actions:

"D5783
(Rev. 166, Issued: 02-03-17, Effective: 03-03-17, Implementation: 03-03-17)
§493.1282 Standard: Corrective actions
(b)(2) Results of control or calibration materials, or both, fail to meet the laboratory’s established criteria for acceptability. All patient test results obtained in the unacceptable test run and since the last acceptable test run must be evaluated to determine if patient test results have been adversely affected. The laboratory must take the corrective action necessary to ensure the reporting of accurate and reliable patient test results.

"Interpretive Guidelines §493.1282(b)(2)
When an internal control fails to fall within the defined limits of acceptability, the laboratory must identify the reason for the failure and correct the problem before resuming testing of patients. The laboratory must evaluate all patients test results since the last acceptable external control."

Notice that CLIA says "evaluated to determine if patient test rulest have been adversely affected." So the lab gets to decide if patient specimens were impacted, and only those impacted will then have to be retested. In the interpretive Guidelines, the phrase "evaluate all patient results" is ambiguous. Evaluating does not automatically mean retesting.

Here then, are the real-world questions:

 "1. We usually run QC after doing maintenance (daily, weekly and monthly). So, if we run QC after maintenance and if QC were out, do we need to do patient look back or rerun previously run patients? Or if part of the troubleshooting we did was to calibrate the analytes to make QC within range, do we need to rerun previous patients? Or only if the violation is 1-3S, 2-2S, or 2 of 3-2s?

"2. We run QC every 8 hours on some analytes based on sigma. So, when we run every 8 Hr QC and if it is out and we need to calibrate it, do we rerun previous patients that were run before the QC went out of range? If we just rerun using fresh QC bottle, do we also have to rerun previous patients?

"3. If during monthly review, I noticed some issues like imprecision, and we called the tech support and engineer saw something wrong with the instrument and did some fix, do we need to rerun previously reported patients?

"4. Also if we have QC out of control and we need engineer to come in to fix it and engineer found something in the instrument and fix it. Then do we need to do patient look back or rerun previously resulted samples?"

We don't need to address the specifics of every nuance of the question. Let's begin by positing that we've optimized our system to eliminate as much as possible any false rejection, and we've narrowed our use of QC rules to the necessary few. Thus, when the QC says the run is out-of-control, we're confident this is a medically important error that is occurring.

[Note that this assumption would eliminate some of the issues described in the questions. If there are "errors" that are not large, not of the size that are medically important, we don't flag, we don't stop the run, those patient samples are fine. ]

If we have a real error there are a number of re-testing scenarios that can be explored:

1. Re-test everything. This is the safest response. And the most time-consuming and expensive one.

2. "Walk back" the re-testing. Test the last 10 patient samples of the run. If there was no significant difference in the re-tested samples from the originals, assume that the patients were not impacted by the error and call it a day. If any of those 10 re-tested samples were significantly different, then re-test the next previous 10 patient samples. Lather, rinse, repeat as necessary. Notice that the use of the round number 10 denotes that this is a rule of thumb, not anything truly scientific.

3. Retest only abnormal patient samples.

4. Retest only patient samples that were close to a decision limit.

5. Retest only patient samples that differed significantly from the values of previously tested samples.

 There are certainly other rationalizations that can be employed to reduce the re-testing burden.

Bear in mind, the regulations are not clear about how much re-testing should be done. Whatever policy is adopted by the laboratory must be documented, and some attempt at formulating the rationale should be stated. Then, at least, an inspector can match your practice to your policy.