Search the Elsmar Cove!
**Search ALL of Elsmar.com** with DuckDuckGo Especially for content not in the forum
Such as files in the Cove "Members" Directory
Social Distancing - It's not just YOUR life - It's ALL of OUR lives!
Me <——————— 6 Feet ———————-> You

Heated Sealed Packages - Sample Size for OQ (Operational Qualification) and PQ

L

Learning Bob

#1
Hello,

I'm trying to determine sample size for my OQ run for heated sealed packages. I plan on running 3 different experiments to find my center of the process window and what my outer limits are. I plan on performing Peel testing and Dye penetration testing. I don't know if I need 30 samples for each experiment, 500 samples for each experiment and so on? Also the dye penetration testing will be pass/fail (attribute data) and peel testing to be measured in lbs (variable data).

I hear a lot about minimum sample size can be 30??

This is for a new heat seal machine and for a new FDA product as well that will go through ETO sterilization.

Also, any help for sample sizes for a PQ would be appreciated.

Thanks
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#2
The "minimum sample size" of 30 is a rule of thumb for continuous data only. If you know nothing about a process a sample size of 30 can get you started.

for OQ, however, it is probably too big or too small...

For variables data it will be important to understand the process standard deviation (how? Do you have past data for similar processes?, do you have any development data on the new equipment? You can always use 1/5 the tolerance spread as a first guess), how precise you need to be (+/- Delta) and the confidence level - traditionally (mythological lay) the confidence is set at 95%. Then you can use the formula: n = [1.96*SD/Delta]^2 note that delta (the amount of precision) must be in the units of measure of the characteristic, NOT a percentage...

For pass/fail data the sample size is based on the defect rate that is either ACCEPTABLE or NOT ACCEPTABLE. This will give you the formulas to use. Using an ACCEPTABLE defect rate will yield a smaller sample size than if you select a NOT ACCEPTABLE rate. There are several formulas so it would be helpful to understand what defect rate you are interested in. you can say 0 but the sample size will VERY large.
 
L

Learning Bob

#3
Hi Bev,

Is there a good book or website where I can learn more about sample size for OQ/PQ?

It sounds like there can be many different sample sizes and not a "black & white" rule.

Thoughts?:bigwave:

Thanks.
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#4
hmmm. I don't know of any single book that addresses IQ/OQ/PQ directly.
I'm not aware of any single book that adequately covers sample size determination either. some will have most of the relevant formulas, but you must still apply your knowledge of the process and it's physics to select the correct formula.

I am in a training course right now - as the instructor - but in a few days I'll try to post a paper for my folks that deals with determining sample size.

You might try to get a copy of the following books:
Moen, Ronald D., Nolan, Thomas, W., Provost, Lloyd P., “Quality Improvement through Planned Experimentation” 2nd Edition, McGraw-Hill, 1999. This will cover various experimental design structures that will apply to OQ.

Sleeper, Andrew, Design for Six Sigma Statistics, McGraw-Hill, 2006 This is an expensive book but it has most of the various sample size formulas in it...not sure if it's worth it as there are also some misleading things in it, but I have found it handy as 'look up' reference...

Also this web site has nice brief descriptions of various formulas and approaches and its free:
NIST Engineering Statistics Handbook, NIST/SEMATECH e-Handbook of Statistical Methods
 
U

uhohraggy

#5
The sample size for the OQ doesn't really matter. The IQ and OQ are really about the equipment. For example... a heat sealer typically has 3 different parameters, temp, time and pressure. Let's just take temp, and say it operates from 30C - 60 C. You might take 5 samples at 30, 40, 50 and 60 C each, but you would also test each temp point with a thermometer. The peel force at each of those with standard deviations gives you an idea of the value and variation you get.


Here are a couple of links you might find useful
Common Mistakes in Validating Package Systems


https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0ahUKEwjKqLmc2PzSAhVHi1QKHVgACuYQFggcMAA&url=http://www.imdrf.org/docs/ghtf/final/sg3/technical-docs/ghtf-sg3-n99-10-2004-qms-process-guidance-04010.pdf&usg=AFQjCNGkjgG0uA5eX75IylBj4v_5akiMEQ&sig2=fVWdhiytj-T9J-zOJkR8rA&cad=rja
 
Last edited by a moderator:
U

uhohraggy

#6
Oh...and on the PQ... they will be different for dye penetration testing and peel force, but both have to be based on risk.

Peel force you need to know the expected value and variation (as well as confidence and reliability). For dye penetration, what is the assurance level you need, again....confidence and reliability.

For example...if you test 150 you can say you have 95% confidence that your failure rate will be below 2.4%, and that is with 0 failures. Is that good enough for your product? What's the distribution cycle? Is it sterile? What's the risk to the patient? Lots of factors to take into account.

See

http://www.mddionline.com/article/assessing-passfail-testing-when-there-are-no-failures-assess
 
#7
"The sample size for the OQ doesn't really matter. The IQ and OQ are really about the equipment."
I really agree with you, then where can I check the rationale for sample size of OQ
 
Top Bottom