US regulation of food ingredients and supplements employs a concept known as, “generally recognized as safe,” or GRAS. Under FDA regulation a food additive or substance is GRAS if qualified experts believe it is safe based upon scientific evidence. However, for substances in use prior to 1958 GRAS does not require any scientific evidence; “a substantial history of consumption” is sufficient.

The Dietary Supplement Health and Education Act (DSHEA) of 1994 allows for herbs and dietary supplements to be regulated for safety as if they were food, and therefore GRAS applies. (It also allows for so-called “structure-function” health claims without any requirement of evidence.)

The premise of this part of the GRAS rule is that if a large number of people use a substance over a long period of time, any safety issues would emerge and would be known. This premise, however, is naïve, and is contradicted by historical evidence.

Marketers often rely on the naturalistic fallacy to sell the safety of their supplements. “All natural,” a term without unambiguous definition or legal regulation, is almost ubiquitous on supplement advertising. This is little more than the naturalistic fallacy, however. Being “natural” (whatever that actually means) is no guarantee of safety. Plants and animals evolved a wide variety of toxins and poisons for their own purposes. I would not recommend eating a random plant unless you know exactly what it is – most “natural” things will kill you or at least make you sick.

GRAS, however, is essentially an appeal to anecdotal evidence, specifically stating that a large amount of anecdotal evidence is reassuring. It is not, however. Anecdotal evidence is fairly reliable when it comes to immediate dramatic effects. If a substance rapidly makes anyone who eats it violently ill, people will figure that out without having to do blinded scientific studies. All cultures identified local edible plants, poisons, and even some basic drugs like hallucinogens through simple trial and error. GRAS is therefore adequate for asserting that you won’t drop dead from the product. Although, experience with the supplement ephedra indicates that even this might not be enough, if the percentage of people who drop dead is low. The FDA received 37 reports of sudden death in people taking ephedra supplements, and many more reports of serious side effects. Ephedra is a stimulant that raises blood pressure and puts strain on the heart. Still, it took an FDA ban in 2004 to remove ephedra from the supplement market (although not from traditional Chinese herbal remedies, so beware).

What about more subtle but serious side effects? What if an additive or supplement increases the risk of cancer, or heart disease, or if it causes liver or kidney damage in a small percentage of users? These are the kinds of serious long term side effects that every drug is analyzed for in clinical trials. Even then, some drugs slip through and the increased risk is only detected in after-market monitoring (when many more people are being exposed to the drug).

The question is – if a supplement caused any of the above risks, would anecdotal experience be sufficient to detect it? The answer, I think, is a fairly clear, no! It takes rigorous data gathering and statistical analysis to detect an increased risk in a population, isolating one variable like a drug or supplement.

If a 55 year old man with high risk factors for heart disease has a heart attack, that does not sound any alarm bells. The fact that this person’s risk for heart attack was increased by taking the drug Vioxx, for example, was not detected until large studies identified the risk. The exact same risk as Vioxx from an herbal supplement would probably be undetected.

Now that the supplement market has exploded (thanks to DSHEA) and there is increased interest in studying herbal supplements, it is no surprise that the same kinds of side effects seen in drug research are emerging. Herbs, after all, are drugs.

A recent report of three cases documents acute liver injury from taking weight-loss supplements, such as Hydroxycut from Herbalife. What would we find if this product were studied systematically? These are not isolated cases. A study found that 18% of reported liver damage cases from 2003 to 2011 were associated with the use of dietary supplements.

Liver damage has been specifically linked to kava kava, black kohash, saw palmetto, valerian, and other herbs.

The Chinese herbal remedy aristolochic acid has now been linked to kidney failure.

In addition to health risks from the herbs themselves, most dietary supplements contain contaminants (such as heavy metals – at least they are all natural contaminants), many at dangerous levels.


The concepts of “natural” and GRAS are not adequate to establish the safety of supplements, especially those that are intended to be taken daily over a long period of time. The more these products are studied the more it is found that they contain the same array of common health risks as drugs. The evidence strongly suggests that the regulations should be altered to provide for a similar requirement for scientific studies testing the safety of dietary supplements before they are allowed on the market. After market monitoring, as with drugs, should also be required.

If a substance can have a positive health effect in the body, then it can have a negative health effect. Nothing is risk free.

None of this even addresses the issue of efficacy – do the herbal products have the health benefits with which they are marketed? The results of large clinical trials of the most popular herbal products suggests that the answer is largely no.

Therefore we have a class of health products with unknown or no health benefits and largely unknown but emerging health risks. This is not a rational risk-benefit ratio for a health product widely marketed to the public with very permissive regulation.

Steven Novella, M.D. is the JREF's Senior Fellow and Director of the JREF’s Science-Based Medicine project.