Hi Brian, I'll do my best!
1. Do you have an expectation from your suppliers that they provide an allergen-free product? (Not exactly. Our Raw Material Specification states in the Allergen section: "Does not inherently contain any allergens. Allergens can only be introduced by cross-contamination with other grains that contain allergens. Ensure that the supplier guarantees the ingredient has been isolated from any wheat or soy (raw grain, products or by-products)." So, while we expect that there is care taken to guarantee isolation, there is no requirement for cleaning common equipment like combines, bins, augers, etc.).
This is a great specification. As you move into human food, include the 20ppm limit so that if you have a lot that fails, you can reject it on the basis of your specification.
If this is so, then you would want to test product samples from say, one shipment a year from each supplier (maybe during harvesting season of allergens of concern, where you might be more afraid of contamination). This helps you confirm that you are receiving an allergen free product with data, and holds your suppliers accountable to their own allergen programs. Sample testing of that type of product requires a blender, and you can run each test for 30-40 bucks in-house if you wanted to (just accept the liability of being the tester, and some customers may require it to be 3rd party). (I think this may be our best bet, because even if there is minimal cross-contamination, it should still produce low enough results to come in under the 20 ppm suggested limit).
Be careful with your assumptions. 20ppm is not as high as you may think (see link for visual). On your failures, you may not even see the contamination. Allergen verification is tricky business and very easy to fail. Because your process has not been validated to remove allergens (as discussed below), your corrective action would be to re-route the lot to a non-human use (probably your easiest rework based on current business model) or reject the lot entirely. Most cheap allergen tests aren't quantitative either, so you'll get a pass/fail reading based on a limit, check the limit of the test and make sure you're not using an extra sensitive test, or one that isn't sensitive enough. That being said, I don't work with raw grains, and I know the standards are going to be different at that level. My best guess is that those upstream processors sell the grains from "transitions" from one silo to another probably sell them first for animal use, then once likelihood of cross contamination is low, sell that section for human use.
2. If not the suppliers, do you have an expectation that your manufacturing process creates an allergen free product? (Our manufacturing process should, theoretically, reduce any concentration levels of allergens that may have been present in the raw material. We utilize scalpers, screeners, aspirators, etc. - each step should (again, theoretically) remove some of any present allergens. How much? We don't know. But our process, at the very minimum, would do nothing to increase the risk).
I agree, as long as you remain a single-allergen facility, and your process is not designed to remove allergens, then your allergen program remains exclusively supplier-based. If this ever changes, and you use your setup to remove potential allergens, then the program needs to change, and you need to have a strict cross-contamination prevention program between raw and processed product. Essentially, treat incoming grains like raw beef, and outgoing grains as cooked beef, keeping equipment, tools, and personnel separated (within reason).
This is even easier, and the interval depends on you. Simply come up with a way to "test" your process by introducing an appropriate amount of oats with soybeans contaminating them into your screening and other steps, and then testing the product that comes out the other side. This should happen at minimum annually or whenever you change the equipment. But allows you to do very little testing overall while providing a scientific evaluation of your equipment. (While I like this idea in theory, I don't know how feasible it is in the real world for our facility. We have a dry system, which is designed to primarily be self-cleaning via product flushing. This would potentially contaminate a very large amount of product that follows through the system, or else we'd have to thoroughly clean the system in a way that it's really not designed to be done. I have to think about this one some more...)
This is inherently the problem with challenge studies, and the solution is generally to use a surrogate. If this was a micro-validation, I would use a non-pathogenic bacteria rather than introducing e. coli to my equipment. You may be able to do this with a non-USA allergen like mustard seed, but again that wouldn't be a very good validation because mustard seed would be handled differently from soy contamination. The best way to do with would be to use the allergen of concern, but do it while you're doing a long run of livestock products, so that if you failed the validation and contaminated your equipment you can "flush" it with product that can still be sold without meeting the allergen standard.
This would be a great minimum start to a program to give you confidence before moving into these new markets, but remember, you cannot test your way to food safety. Validating your suppliers, processes, and testing at an interval that provides confidence (remember, if 5% of product is contaminated you only have a 1/20 chance of seeing it on each test!), is what you'll need to make sure you provide safe product. It's unfortunate that grains all got exemptions to basically all of the FSMA final rules. Good luck! (Thanks! But what do you mean, it's unfortunate that grains all got exemptions? I don't recall reading or hearing about that...what exactly do you mean, and where in the final rules is it located?)
Grains in the "raw" state essentially got a free pass, placing the burden on grain processors like yourself, which makes supplier approval and compliance enforcement more difficult for raw ag.
The entire FSMA produce rule, in the exemptions section: "The rule does not apply to: Food grains, including barley, dent- or flint-corn, sorghum, oats, rice, rye, wheat, amaranth, quinoa, buckwheat, and oilseeds (e.g. cotton seed, flax seed, rapeseed, soybean, and sunflower seed)" http://www.fda.gov/F...A/ucm334114.htm
Exemptions from food transportation rule: "Transportation activities performed by a farm"http://www.fda.gov/F...A/ucm383763.htm
Exempt from preventative controls: "The definition of a ‘farm’ is clarified to cover two types of farm operations. Operations defined as farms are not subject to the preventive controls rule...Secondary Activities Farm: This is an operation not located on the Primary Production Farm that is devoted to harvesting, packing and/or holding raw agricultural commodities. It must be majority owned by the Primary Production Farm that supplies the majority of the raw agricultural commodities harvested, packed, or held by the Secondary Activities Farm." http://www.fda.gov/F...A/ucm334115.htm
Earth2O, you definitely have helped me think about this allergen control program in a new way. I had been thinking purely in terms of product testing, because of my background...but this approach seems so much more appropriate. Thanks again!!
Remember that even though you may consider these verification events critical to food safety, because you have no control in place it would not be a CCP. This is a prerequisite program in which you approve suppliers and verify they are meeting your raw material specification at some interval using testing. I would also encourage you to make that testing as meaningful as possible by taking it directly from the transport vehicle, targeting seasons where multiple crops may be stored or harvested on shared vehicles and containers, and grabbing from the beginning of the lot, where leftovers from the previous product are most likely to be included, or in sweepings/fines where non-whole seeds may collect and act as a source for contamination. You could also go crazy and instead test the fines from your screeners and such, and establish a threshold amount of allergen containing material that is collected and concentrated there, but I have no idea what limits would be or how indicative that might be.
Also, don't get trapped in thinking that among an entire lot of product (so many tons of product) the concentration is <20ppm. That doesn't matter. What matters is whether a single consumer might eat a single meal with a higher concentration. If I'm allergic to peanuts, and throw one peanut into 5 boxes of hamburger helper in a lot of 1,000,000 boxes, the concentration for the lot is well below 1ppm. But for the consumers who get those 5 boxes, the concentration is much higher, and your lot statistics don't matter.
Best of luck with approaching this! Remember to keep scope small, risk assess, and take baby steps with implementation. And absolutely critical, don't test for curiosity alone. Never test product unless you have an agreed upon corrective action for a positive result. You cannot unlearn a positive result or test it away, so be prepared to rework, recall, or destroy any product you test.