Packaging / Heat Sealer Validation - IOQ (Installation/Operational Qualification)

[Sam Lazzara]

I would not "require" Cpk analysis, and none of the standards I am aware of have any requirements for Cpk or a particular type of data (variables/attributes). Applying Cpk or statistical tolerance limits would be a sensible way to state your specifications, at least for those that are based on variables data. Caution: Be careful not to automatically assume your data is normally distributed.

Some of the tests (like bubble leakage and dye penetration, for example), are attribute (pass/fail) tests so variables analysis is not possible. For attribute data you could have specifications for confidence/reliability such as 95/99, 95/95, 95/90 which would require no failures for sample sizes of 59, 29 and 22, respectively (binomial distribution stats).

 

[rick76]

Given that I’m in the same situation on validating a new sealer, I was wondering if someone can share a IQ/OQ/PQ validation report for band sealer to seal tyvek pouches, we are in the process to purchase a new sealer therefore i need assitance what to look for(medical validated sealer or non-medical sealer. In my case, The entire product required sterilization, however some are ETO and others are Gamma sterilized using the same sealing process. Please advise if you have a template that includes testing parameters that are similar to Sam and Wes77 requirements. Also during the validation What are the parameters that need verification and calibration?

 

[alex.Kennedy]

I do not know where you got that information from but it is completely erroneous. The regulators treat packaging and labelling very seriously and require it to be fully validated. We have just completed a contract for validating six packaging machines that originated from a very concise and detailed 483.

Alex Kennedy

 

[gdwaikle]

Hi,

In response to the last post- The reason why wouldn't you be concerned about a product pachage that is not steriilized or distributed is that your main concern for product in that case is making sure it gets to the end user without failing out of the package. Whereas packages that are used as sterile barriers are trying to insure there is are no human patients who will get an infection and die.

Right now this topic is very hot for the FDA who seems to be very concerned about process validations of heat sealers. One of our suppliers just had an FDA inspection and the primary focus was heat sealer validations.

With regard to the reason for using 3 production lots to prove the process is stable - Runnng only one or even two lots would not give you enough normal variabiliity to prove the process is stable. When you get to three lots you have some assurance that between shift changes and personnel changes most of the normal variabilities that occur have occured. At that point you have some assurance that your process is stable and in control.

I can tell you there is a some expectation by the FDA to see 3 separate runs.

 

[SylvieS]

Hello, how would you address the sampling size determination based on % confidence and reliability when having multiple cavities? Just act as 1 cavity 1 validation or doing a variation analysis between cavities during OQ - but then again on how many samples would you do the variation analysis ?

 

[gdwaikle]

Hello, how would you address the sampling size determination based on % confidence and reliability when having multiple cavities? Just act as 1 cavity 1 validation or doing a variation analysis between cavities during OQ - but then again on how many samples would you do the variation analysis ?

Sylvie, as you are likely aware if there are multiple cavities, each one represents one output of the mold. Each one has to be treated as an independent machine so you should take samples from each one. However, if in your OQ you discovered one o more of the cavities has more variability than the rest you might be able to sample those as worst case.
I will give you an example of one of my experiences with molded plastic parts from mold with many cavities The parts in question were two mating parts that were part of a sample collection sub-assembly. We began receiving field complaints which were caused by this component. The two parts had taper fits that when put together would be connected allowing a donors plasma to be collected. The parts started coming apart, which we began calling a spontaneous disconnection and allow the donors blood to flow through the bad connection. Luckily no donors were injured.
In our investigation for the root cause we found out that the mold being used had severely degraded and the angle of the taper on several of the cavities were creating bad parts, which nearly crippled our companies ability to function.
The bottom line is how many samples you take to perform the PQ may be as many as 30 per cavity or less if you perform a statistical study first to limit the samples you need to take. Warning though you really need to pay attention the make sure each cavity is identified on the parts in case you need to do an investigation later. And you really should perform periodic revalidation of the mold.