What parameters would you cover in the validation of an analytical method used only in assesment of clenliness of an manufacturing device?

I'm validating a filtration unit where we have to flush out a toxic filter preservetive before the system can be used. I got an analytical method from the supplier to detect this preservative. I couldn't get any method validation documentation from the supplier and I think I have to do some amount of validation/qualification for the method, myself. I was thinking atleast: specificity and LOQ. What about repeatability, linearity, accuracy, robustness and intermediate precision?

ICH divides analytical methods to three categories: identification, testing for impurities (quantitative/limit) and assays. In what category this would fall in?

What you are doing is essentially a limit test.

I have question for anonymous above (June 5th): As I understand it a 'limit test' is when you don't focus on quantifying something but, instead, just ascertain that it is below a certain level. What is the reason for taking this different approach in some cases. Why not just evaluate the result numerically against a spec., as would be done for most situations?

Thats exactly what I've been thinking as well. My reason would be that I don't need to do so much validation for a limit test. If you have a look at the table on ICH guideline Q2A on page 4:

http://www.emea.eu.int/pdfs/human/ich/038195en.pdf

you can see that quantitative impurity test needs full validation whereas limit test requires only specificity and detection limit?? What is the rationale, you can say you have a limit test??

A limit test is more of a qualitative test - is it there or isn't it. It gives an answer of > than a certain level or "None Detected".

A quantiative limit test gives you an actual quantity above the LOD/LOQ, or None Detected. Depending on the range over which you validate the method you may also need to set an upper limit for the method, so if you get a result above the range you have to report "Greater than X" because you can't acurately say that the value reported is a true value. This value must be well above the passing limit for you test.

Another way of looking at this: a limit test is appropriate when the situation calls for "go / no go" (i.e., the only acceptable level is *zero*). An extreme example might be looking for product tampering (remember the "cyanide in the tylenol" case from a couple of decades ago? http://www.snopes.com/horrors/poison/tylenol.htm). You don't need to know *how much* is there, there shouldn't be *any*!

Cleaning validation falls in the same category.


-- Anonymous the second

You are trying to study assay for residual substance on a manufacturing device. Am i wrong ? If im not, its cleaning validation. and you need to :
- validate your method and determine LOD / LOQ
- study recovery (from the same material with the device ) in lab. conditions. For this, you may put your substance solution ( includes known concentration of subs.) on a measured part of a plate.Wait for a few hours ( depends how many hours it waits before cleaning in manufacturing area )Then swab the surface and analyse it. You will need this result to calculate real resudial value.
- After cleaning of the device you should get swab samples as you did before in lab.and you should repeat this process for three times.I mean after 3 cleaning, 3 sampling

actually there are 2 more sampling method for cleaning validation . You may do rinse sampling or placebo sampling as well.but swab seems more sensitive.

hope helps...