Sample Size rationale to reduce size and use special levels

[Vetty007]

Hello everyone,

I have a question regarding the sample size of an in-process control and the resulting batch release. Since the product is just a folded, dry wipe in a sachet (low risk overall, deviations in the folding or in the size are hardly relevant to safety), that has been in production for more than 15 years - sample size was always 50 pieces - I would like to stay with such a low sample size, but at least be able to justify it accordingly or slightly adapt it.
According to ISO 2895, I would have to take 500 samples for a batch size of 35.500 pieces, if I take the batch size range below (because I only slightly exceed this), I still have to pick 200 samples. If I could use one of the special levels as a basis here, that would be great. Unfortunately, I can hardly find a definition of when and to what extent they can be used. Can anyone give me a hint, where I can finde more details/definitions?

In my case, I am thinking, if its possible to check 200 or 500 pieces, for example, for integrity of the packaging, proper printing etc. and then add these samples back to the batch (assuming they are in spec) and then use just a part of them and with an amount according to the special level, to open the packs and check the content? Until now we took 50 samples, open them and check if the wipe size is correct and folded properly. Checking the integrity of the packaging and printing is included and done before opening, but as this could be easily done and the samples used afterwards (if allowed to do so?) checking 200 would be acceptable.
Even if the value of the larger number of samples is not that high, it is still annoying when it ends up in the garbage afterwards, esp as there are actually hardly any errors here and, above all, no relevant ones. In this respect, it is certainly acceptable to reduce the sample size, the main question is according to which specifications and to what extent.

I would be happy to get some feedback about my thoughts for testing or new ideas, where to look for more information or suggestions or literature recommendations.....

Thanks a lot in advance

 

[Mike S.]

Your sampling plan should explain when using special levels is appropriate.

Have you considered C=0 sampling plans? Have a look at ASQ H1331 by Nicholas Squeglia.

 

[Vetty007]

Thank you for your comment and recommendation. I am not sure, if the ASQ will be accepted by an EU NB, our NB likes to see german standards or ISO andthus I haven't considerd it yet. . Any experience with the acceptance of ASQ by EU NB?

You are right with the comment but its esp. part of my question, how to explain the use of these special levels.

 

[Cybel]

I don’t have experience to give you an answer, but I would like to share some thoughts.

As stated in ISO 2859, special levels "may be used where relatively small sample sizes are necessary and larger sampling risks can be tolerated".

I guess the value of “necessary” may be not objective enough (for example: at what point, exactly, the inspection cost becomes too high compared with the risk and final cost of the device?), so I would start evaluating the risk of a larger sampling in my risk analysis, bearing in mind that inspection is, for sure, a protective measure in the manufacturing process but not the only risk reduction measure. Maybe you can increase control over the processes such as printing, cutting, folding or others, for example:

- what is the risk if the wipe size is incorrect?
- can you increase production methods to ensure appropriate wipe size?
- can you validate this process?

(My subsequent and collateral question: if process validation must be performed in case the results of that process cannot be fully verified by subsequent inspection and test, could process validation also be performed in case I don’t want to verify the results of that process, for a number of reason, so that I can avoid sample in process and final controls?)

 

[Zero_yield]

Seconding Cybel here. This sounds like a job for validation.

Do you already have other in-process controls, like an ERP system, to ensure that the correct materials end up in the correct batches?

Questions to ask yourself:

1. How frequently are is this control finding deviations that:
...could affect patient safety?
...could have regulatory impact?
...could cause an unhappy customer or impact the business?

2. How often do we have a complaint that this control should have caught?

From there, you should be able to do some risk analysis about what you're getting out of the current control.

 

[Mike S.]

Thank you for your comment and recommendation. I am not sure, if the ASQ will be accepted by an EU NB, our NB likes to see german standards or ISO andthus I haven't considerd it yet. . Any experience with the acceptance of ASQ by EU NB?

You are right with the comment but its esp. part of my question, how to explain the use of these special levels.

Then maybe consider ISO 28593: 2017 .

You said the risks were low.

Is your process in control?

Seems like you should be able to do a risk analysis and come up with options.

 

[Vetty007]

Thank you for all the valuable comments and suggestions

We already controll several parameter during and after production for the (partly) finished product, thus I was thinking to splitt the required 200 test samples, using 50 for a destroying testing (the value in term of cost is very low and not a good reason for a low sample size) and the other ones for the other tests and also "re-use" e.g. the final packages on which we check the correct printing, sealing and packaging of the correct amount of wipes sachets and as this is only a visuable testing, it should be possible to add them to the batch (provided they are in spec) afterwards. We have done a process verification and validation and reached good results (Cpk minimum 2) and of course we have a ERP and make sure to only use the correct material. Thus in general the risk is very low, and mostly without potential influence on patient safety, customer satisfaction or even regulatory aspects. The only risk is a microbial contamination, but the material is checked, the sachet has an antimicrobial coating and the machines were disinfected and checked for contamination before use too - picking 50 sachets at the end of the production line should be enought to check the finished product for contamination. I think I wouldn't finde more positive samples if I collect more - if there is a problem with contamination the 50 samples should be enought to find this problem, if not, more samples will not be helpful .....

Until now we didn't had a relevant complaint, it was just peanuts like, that e.g. the LOT printing wasn't placed correctly, or an arrow to show where the sachet should be opend was missing- All of this is now taken into account in our controlls and it was easy to correct these things.

Okay, then I will buy this ISO 28593: 2017 - compared of one the other ISO with over 200 EUR its cheap, but hopefully includes helpful content.