Microbiology Sampling Plan
Hi All,
I'm yet another person looking for guidance on microbiological sampling plans...I've done my best to review previous topics on the board but I still don't have a solid grasp on how to design a representative system. I'll hash out some info about our process:
We are a dietary supplement facility focused on freeze-drying plants to create powders. These powders are then bottled as loose powder (ie for smoothies), or encapsulated into HPMC capsules. We currently have 3 points of sampling:
- Raw Material delivered to plant
- Batch of freeze-dried powder after the drying process
- Finished product, bottled and labeled
Currently we take one 50g sample at each of these points, for each lot. In other words, each lot of raw material, in-process material, and FG gets one sample. Of course, this presents an issue because the sample size does not vary depending on the batch size. And as always, a big challenge for us is the lack of resources to significantly increase our testing.
We test:
APC, Coliform, Ecoli, Yeast, Mold, and Salmonella on these samples. Soon, for the powders we will test water activity. Sal/Ecoli have zero tolerance for detection in our product, while APC/Coli/YM are based on spec limits.
To increase testing but stay within our means, I'm thinking of increasing the sample size of in-process material and raw-material testing based on poundage of material. FDA standards require that representative samples are being collected (111.105[f]). Given that the number of samples would vary based on batch size, is this compliant? For example, if 1 sample is collected per 100lbs of freeze-dried material, we'd collect 3 samples from 300lb run. Guidance on compositing these samples would be appreciated too...if we could composite the three samples as one test that would be great.
Hi Brothbro,
You may need to study the lot sampling/estimation aspects of (a)(APC, Coliform, Yeast, Mould) and (b)(E.coli, Salmonella) which are rather different in respect to the final objectives.
I daresay you appreciate that for estimating average micro "quality" values using small samples / random sampling the batch size is theoretically, approx. insignificant. It is the absolute number/quantity of samples which is critical. And similarly for detecting zero-tolerant species.
3 random samples/lot is indubitably better than 1 but for zero-tolerant items likely present in small percentages, negative results are highly likely and of limited value. Of course you might get "Lucky !".
FDA's BAM offers detailed info on sampling/compositing for Salmonella. Sadly, internal factory labs cannot usually approach their recommended sampling densities.
Hi Brothbro,
You may need to study the lot sampling/estimation aspects of (a)(APC, Coliform, Yeast, Mould) and (b)(E.coli, Salmonella) which are rather different in respect to the final objectives.
I daresay you appreciate that for estimating average micro "quality" values using small samples / random sampling the batch size is theoretically, approx. insignificant. It is the absolute number/quantity of samples which is critical. And similarly for detecting zero-tolerant species.
3 random samples/lot is indubitably better than 1 but for zero-tolerant items likely present in small percentages, negative results are highly likely and of limited value. Of course you might get "Lucky !".
FDA's BAM offers detailed info on sampling/compositing for Salmonella. Sadly, internal factory labs cannot usually approach their recommended sampling densities.
Hi Charles, thanks for the response. With stats it seems you can never design a system that works with absolute certainty, only one that operates with a tolerable amount of risk. At least as far as I can tell...
I've found a white paper from ASTA that is relevant to my industry, I'll link it here for others who might like it. As far as sampling plan literature goes, this one is pretty concise and easy for a beginner like me to understand!
Hi Charles, thanks for the response. With stats it seems you can never design a system that works with absolute certainty, only one that operates with a tolerable amount of risk. At least as far as I can tell...
I've found a white paper from ASTA that is relevant to my industry, I'll link it here for others who might like it. As far as sampling plan literature goes, this one is pretty concise and easy for a beginner like me to understand!
Hi Brothbro,
Thks for the informative document although it is somewhat antique (no refs post 2008, majority of links broken [not so unusual :smile: ]).
With respect to Salmonella, BAM is more likely of current applicability, for example with respect to sample sizes and compositing multiple samples for Salmonella analyses.
The ASTA website is definitely a Goldmine for the Spice Industry although many interesting-looking items are sadly "member-only".