ISO 17025 Validation Procedure - Need Example

T

ThomasMichael

#1
Anybody has an example/sample method validation procedure compliant to ISO 17025:2005?
Thank you in advance.
 

Hershal

Metrologist-Auditor
Staff member
Super Moderator
#2
Validation under ISO/IEC 17025:2005 is not as well defined as perhaps it could be.....I would begin by stating what the procedure is, what it is for, consummables, equipment, etc......and what you expect it to accomplish.....have management buy off on it.....

Run the procedure or software or whatever.....record and report all the bugs and glitches you find.....fix the important ones.....run a regression analysis to make sure the fixes did not make anything else go bad.....

Provide a report to management clearly identifying what the expected outcome was and what the actual outcome was, and the two must be relevent to each other.....

You will need a statement that the procedure or software that you are validating is acceptable for intended use.....and again have management buy off on it.....

Keep those records, and hopefully you will be OK.....

HOWEVER, check with your accrediting body to make sure that the approach is acceptable to them.....

Hope this helps.

Hershal
 

Wolf.K

Involved In Discussions
#3
Hi,

This thread is 10 years old, but I wonder if someone has a methods validation procedure at hand?

I am asking, b/c we are setting up a small in-house laboratory, and we noticed that nobody of us has ever validated lab methods. We are all Ph.D.'s (biology, chemistry...) but admittedly, at the university we verified methods only - that is, we set up experiments, and asked our advisors, if the p value was ok for them...

E.g. if there is a method like standard ASTM F1929 "Standard test method for detecting seal leaks in porous medical packaging by dye penetration" - this method does not need to be validated, only verified, if I understand ISO 17025 correctly. On the other hand, IQ, OQ, PQ should be done to verify (vaidate?) that the method gives the correct results. There I wonder, how many samples are required. 30 trials I'd guess? How does a validation/verification report look like for such a standard test?

Truly yours,
Wolf
 
#4
In the world of 17025 the standard would prefer that you are using "standard methods". Those are "Methods published either in
international, regional or national standards, or by reputable technical organizations, or in relevant scientific texts or journals,
or as specified by the manufacturer of the equipment, are recommended. " Those are considered to be vetted, and do not require that you validate them.

If you are creating your own process it needs to be documented and validated. If you look at 17025 7.2.2.1 Note 2 they list recommended validation methods that you "can", not must, choose from.
In the end an assessor will look at your internally developed procedure and see if you have a record that you validated that it works properly and gives valid results. It needs to be "as extensive as necessary". They will trust your technical judgement unless it is obviously poor.
As for what will the report look like, section 7.2.2.4 presents a list of the elements that are required to be found in a validation report. They are very straightforward, be sure to include them all and your audit should pass.
 

Wolf.K

Involved In Discussions
#6
Thanks for the feedback! Writing up a report including the required chapters is easy. I still wonder about the number of replicates, thou...

The FDA presentation states:

"23PrecisionRepeatability●A minimum of 6 determinations at 100% of the test concentration, or●a minimum of 9 determinations covering the specified range for the procedure (e.g., 3 concentrations/3 replicates each) – my preference to cover the assay range. "

But they do not list a reference for this recommendation.

In the NATA technical note is written:

"The repeatability standard deviation, variance, probability distribution function, etc must be determined with
at least 6 degrees of freedom. This can be achieved for example, by analysing 7 times in a series with one
test item (df=6), 4 times in a series with 2 test items (df=6), 3 times in a series with 3 test items (df=6), etc
(Zar, 1974)."

As so often, I miss very specific requirements...

For e.g. F1929, I could use three differnent values (30, 7, 6) for replications. A certified / accredited lab, which performed this test for us (as part of an accelerated aging study), required 10 samples... As this was not for the validation of the test, I assume that I should be fine with 30 replicates by two users each (60 replicates in total)?
 

Top Bottom