3 replies to this topic

[matthewcc]

Hello all,

 

We (QA) noticed that we have sometimes been manufacturing a product first, then updating/approving specifications for the product that we have already manufactured, just before release.

 

Similarly, we have sometimes been testing a product that we have made using a certain method, then updating/approving the specification to require the method that we have already used, just before release.

 

This appears to me to be backwards in my QA-oriented mind.  The way I read them, GMP's instruct that we write specifications first, then test to determine whether the specifications have been met.  See for example 21 CFR 111.70 and 21 CFR 111.75©. [https://www.ecfr.gov...ection-111.70, 21 CFR 111.75©]

 

That makes sense to me because you first establish the requirements (specifications), then manufacture the product and test it to determine whether the specifications are met.  From a regulatory standpoint, would this (the first two paragraphs, what we are sometimes doing) be viewed as noncompliant in the sense that it is kind of like writing the specs to make it so you can comply with the specs?  Or, is this not important?  (I seem to remember our laboratory defending the way we sometimes do this, saying we need to make sure a method works before we update the specifications to require the method.)

 

This is for manufacturing dietary supplements in the United States, and we have SQF and NSF GMP certifications.

 

Thank you,

Matthew

[Brothbro]

Hi Matthew,

 

For clarity, is this for routine production? Or is it for new products that you're running for the first time? If these are new products, the people responsible for RD/QA should be working together to set specifications up ahead of the first "live" run of a product. This often involves "test runs" of a product that are not intended for sale.

[matthewcc]

This is for routine production.

[Brothbro]

For routine production, it's not advisable to operate the way your company is. Specifications are set to define what a product "is" at certain steps of the process. By constantly redefining what the specifications are for a product, you essentially have none! A common (but not acceptable) tactic that manufacturers employ is:

 

1) Produce a batch

2) Testing reveals the batch does not meet specifications

3) Specifications are hastily "updated" to allow the batch to be released

 

Does this sound familiar? Instead, the non-conforming batch should have been held for investigation. Further processing or corrective measures can be approved by QA to bring the product within the established spec if possible/safe, but the corrective action should not be to simply change the specification.

 

Of course, specifications are allowed to change, but there is a process to that and it needs to be for the right reasons. As a company grows they may change processing methods, reformulate products, or employ new machinery. Changes to product specifications should be documented and supported with data and/or industry/regulatory guidance. If your records instead show that specs are changing as a response to non-conforming batches, regulators will take issue with that. 

 

If you're having trouble meeting spec on products, your teams should be sitting down and taking a look at your production process. If your original specs weren't made carefully it does make sense that you would constantly deviate from them. But your company's response to that shouldn't be reactionary; instead a proactive approach should be taken to redesign the spec from the bottom up, once and for all (until new circumstances arrive).