Hi all,

I wanted to know what is the general industry practise employed in developing sampling plans for Pharmaceutical excipients. I checked with couple of friends and the answers I got was square root of n + 1, or some percentage like 1 - 10 containers would be 100 % sampling and subsequent increase for additional containers and so on.

But what surprised me was that no one employs ANSI Z 1.4 tables. So my question is can we use the ANSI Z 1.4 in the first place as there wont be those AQL as Acc / reject. I was thinking that the ANSI Z1.4 table could be used just as a guide to select the number of containers to be sampled.

Appreciate if I could get some quick info as I need to answer the auditor tomorrow :cool:

On another note, would square root n + 1 be considered a statistically valid plan ?

Pls help :(

Hi all,

I wanted to know what is the general industry practise employed in developing sampling plans for Pharmaceutical excipients. I checked with couple of friends and the answers I got was square root of n + 1, or some percentage like 1 - 10 containers would be 100 % sampling and subsequent increase for additional containers and so on.

But what surprised me was that no one employs ANSI Z 1.4 tables. So my question is can we use the ANSI Z 1.4 in the first place as there wont be those AQL as Acc / reject. I was thinking that the ANSI Z1.4 table could be used just as a guide to select the number of containers to be sampled.

Appreciate if I could get some quick info as I need to answer the auditor tomorrow :cool:
I'm not in pharmaceuticals, so I have no idea what is standard practice, but I can comment a bit on the statistical issues.

From a statistical perspective, the sample size is pretty much independent of the lot size. If you randomly sample 30 items from a lot of 100 or from a lot of 100,000, your statistical certainly would be about the same either way. When the sample size is almost as large as the lot size (form example 30 out of 36), then the statistical certainly would improve noticeably.

I have heard the "square root of n + 1" rule before. It does have some merit. As I just said, you don't need to increase the sample size to maintain the same certainty, but it might make economic sense to check more items in a bigger lot. The bigger the lot, the more you risk by accepting it, so the more sure you might want to be.

The Z1.4 tables are certainly commonly used in many industries. I should say, however, that they are not based on pure statistics. You can say that " a lot that is no worse than the AQL will usually get accepted" but there is no definite rule like "such lots will be accepted 95% of hte time for normal, level II sampling plans." If you really want to know the statisitics, you need to chweck the actual OC curves and accompaning tables in the back of the standard.


Philosophically, I have come to the conclusion that sampling plans are best used as "insurance'. SPC and fool-proof manufacturing processes are much better for ensuring quality. Sampling is a way to check for gross failures. Perhaps a box got mis-labled. Perhaps somthing broke midway thru a production run. If you want to watch for a change from 0.1% defective to 10% defective, sampling is quick and efficient. IF you want to watch for changes from 1% to 2% defective, sampling will do a very poor job.

Hope that helps. At least it migh serve as a starting point for further discussion.

Tim

Dear Mr. Ajit

I don’t have much idea in pharmaceutical excipient sample handling but, I gathered some information that sample selection can either be done statistically based or arbitrarily chosen and appropriate statistically based sampling plan can be use ANSI/ASQ Z1.9–2008 tables for bulk materials and ANSI/ASQ Z1.4–2008 tables for multiunit or discreet populations.

Thanks and Regards
S. Subramaniam

Thanks Tim and Subbu - I will go ahead with ANSIZ1.4.

Btw, does any one have any position paper for sampling by square root n + 1 ?

Hi

I think it is a Pharmacy domain special purpose sample plan , that you can refer a article :「The Square Root of N Plus One Sampling Rule」by Hewa Saranadasa.

Marvin Tsai

Welcome Marvin and thanks for the reply :bigwave:

Welcome Marvin and thanks for the reply :bigwave:

I can't help you, but now I know what an "excipient" is. :D

ex·cip·i·ent (http://img.tfd.com/hm/GIF/ibreve.gifk-siphttp://img.tfd.com/hm/GIF/prime.gifhttp://img.tfd.com/hm/GIF/emacr.gif-http://img.tfd.com/hm/GIF/schwa.gifnt)n. An inert substance used as a diluent or vehicle for a drug.

[Latin excipiens, excipient-, present participle of excipere, to take out, exclude; see except.]
The American Heritage® Dictionary of the English Language, Fourth Edition copyright ©2000 by Houghton Mifflin Company. Updated in 2003. Published by Houghton Mifflin Company All rights reserved.

I can't help you, but now I know what an "excipient" is. :D

:agree1: