**NOTE: all the information presented under the "FDA cGMP compliance open source project" is not intended to be legal advice, nor is it in any way guaranteed that my interpretation of the statute and system for compliance is going to satisfy any individual FDA inspector or compliance officer. These pages are a synthesis of my own understanding of the regulations and how to achieve compliance.
When manufacturing any dietary supplement (herbal or otherwise), the guidelines on current good manufacturing practices (cGMPs) require that specifications be established for everything that goes into what you're making. This essentially means that you have to define a few things and set a standard that you find acceptable, and then test the relevant material and record the results of that test. The regulations under 21CFR111 tell us what specifications we need - sort of. Let's take a look.
A great outline of all of 21CFR111 is here:
and, for the details on specifications, see section 111.70:
The key pieces are identity (what is it?), purity (for a raw herb, is it all the herb in question? Is there dirt and/or foreign non-herbal matter?), strength and composition (for a raw herb, this is full-strength 100% raw herb), and contamination.
The identity, purity, strength and composition for whole or cut-and-sifted (NOT powders, which require a microscope) can be assessed through organoleptic testing (using sight, smell, feel, and taste) if conducted by a qualified herbalist. The contamination, unfortunately, requires help. This is even if you get a certificate from your vendor telling you that there is an acceptably low level of contamination, if any. This is one of the big pieces that we still have to solve as herbalists: the cost for contamination testing of each batch of raw material can get steep. Is there a way to consolidate results on lots of herbs so all can benefit and/or spread the costs?
Identity specifications are established in an herb monograph, and there should also be a voucher sample of the herb in question, as well as a sample of the cut-and-sifted material if that's how you purchase it. Additionally, details should be included about potential adulterants, if an herb is known to be adulterated (goldenseal and yellowroot for instance). Rejection criteria - meaning, if you find this you mail the herb back to sender - should also be included.
AHPA has begun to collect excellent information on botanical identity, including microscopic identity. You should visit:
Purity specifications can include details on adulterants, but also a way to assess (and reject if necessary) how much filth and foreign matter is present in a sample.
Contamination specifications include:
- microbiological testing. What's an acceptable total bacteria count? What about yeast and mold? Note that in many cases, bacteria, yeast and mold are a normal part of, say, a fresh berry. Without them there would be no wine. But sometimes there's too much. And certain specific bacteria, like E. coli and salmonella, should never be present.
- heavy metal testing. How much arsenic, mercury, lead, cadmium are in the sample? It's tricky to determine what an acceptable level could be for a raw herb. Additionally, there is good evidence that tincturing actually removes heavy metals from the final product (because most, if there are any, remain in the discarded marc).
- pesticide testing. Our argument so far has been that using certified organic herbs shifts the burden of responsibility to USDA, and that we are not required to test for pesticides. We do use only certified organic herbs. This may or may not be enough - the jury is still out. I'd be curious to hear of other's experience here, especially if there has been pushback from FDA even for those who use certified organic raw material.
AHPA has put out great guidance documents to help define these specifications. For an overview, see:
And, for much more details on heavy metals specifically:
The above is good, but the heavy metals specifications refer to daily total maximum intakes, not how much can be in an herb you're planning to use. This is a complex question that has to do with the weigh-to-volume ratio of your tincture(s), the dose, and how much heavy metal is left in the tincture from the original raw herbs. Working backwards from the daily limits, you can make a case for how much should be present in a raw herb. Here is an example of how to make such a case, and includes the final heavy metal limits:
Heavy metals contamination rationale.
Again, the problem isn't necessarily defining the specification, but testing for it. These tests can get quite expensive. There are ways to reduce the amount of testing you do, especially once you have some data to go on - but getting there can be an overwhelming financial burden. We need to find a way to collaborate on these types of tests - to spread the financial burden and somehow, effectively, share results if we want to support small-scale manufacturers.
What follow are a couple of examples: the first is a raw material monograph (the specifications) for Echinacea purpurea root. The second is a specification sheet, where the results of testing are recorded to ensure they comply with the specifications. A voice walk-through is below as well.
|Full document .pdf file|
|Full document .pdf file|
One note: these are "controlled" documents. They must be signed, dated and approved by whomever is in charge of quality control; revision numbers must be advanced if any changes are made; and they should be "secure" (i.e. not easily tampered with or alterable). Filled-out spec sheets must be in hard copy, filled out in ink, and stored. Until this documentation exists, and is signed off at the bottom, you cannot use the herbs for manufacturing and they should be stored in a designated "quarantine" area (blocked off by fencing, walls/doors, or even tape on the floor to distinguish the area from one of active use).
Finally, one cannot necessarily rely on organoleptic testing unless the person conducting the test(s) is qualified to do so. While you don't have to convince me that you know what dry, cut-and-sifted Echinacea purpurea root looks, smells, and tastes like, you will have to convince FDA. Plus, herbalist qualification does raise some interesting questions: could you tell the difference between E. purpurea and E. angustifolia? Perhaps. What about Actaea racemosa and Actaea pachypoda dry, cut-and-sifted rhizome? It's not always so straightforward. Again, AHPA is doing a great job putting together a lot of the specifications and images necessary for herbalists to conduct these tests (see, for instance, the detailed Actaea racemosa monograph), but unless you are a botanist or pharmacognosist with a degree, you will have to prove to FDA that you can competently identify herbs. Essentially, you have to find a way to "test the tester", or qualify the analyst, who is ensuring that the organoleptic specifications are being met. Here are two document that give an example of how one might do this:
Analyst testing method (how we test our testers)
Analyst testing record (where we record the test results)
Ideally, you will build a testing record that shows you can competently identify all the herbs you use. Now, the question remains: how do you know that the sample you're using as a way to certify the herbalist's skill is actually what you think it is? This requires either the purchase of a validated sample, the assistance of a qualified botanist or pharmacognosist, or validation through an external lab test. This can add extra cost. Is there a way for us to centralize this somehow? Could analyst certification be offered at herb conferences?
Any additional training seminars, certificate programs, or herb schools that you can attend (and collect certificates!) serve to bolster your case that you are, indeed, qualified to conduct these tests. Build a strong case, and identity of raw (non-powdered) herbs can be tested using our senses and good reference materials.
So, specifications need to be developed for everything - including packaging and labels - and you have to maintain a record that you checked everything that comes in against those specifications before using material to make any kind of product that will go to the public. This is a big undertaking. In the coming weeks/months we'd like to share as much of what we have as possible. Hopefully these documents can be edited, improved, and custom-tailored to adapt to different individual situations. But the scope of the project has already outgrown this blog: we need a platform that allows for collaborative sharing, posting, and discussion. So next week, I will provide links to a separate website that will hopefully be more interactive and allow us to leverage the excellent experience and advice, as well as offers for help and contributions, that have been coming our way. Thank you for supporting the FDA cGMP open source project and stay tuned!