described as a modifying factor, in that whether identifiable subpopulations are particularly susceptible to harm should always be taken into consideration when setting priorities for review.

The principles for evaluating specific types of information are described above, but some concepts are more global in nature, because they are applicable to all types of data or because they are principles for integrating different types of data that may or may not be consistent.

A critical factor in determining toxicity of a compound in a dietary supplement is not necessarily the ingested amount, but rather the unbound (free) concentration of an active ingredient at its receptor site. Once absorbed, distribution of the ingredient is via the systemic circulation to its receptor site.

Bioavailability (i.e., the rate and extent, or fraction, of delivery of a compound to the systemic circulation) has a significant effect on the concentration achieved. Bioavailability is greatly affected by the composition of the dosage form, first pass metabolism in the intestines and liver, and physiological factors, such as the rate of gastric emptying. Bioavailability and the rates of metabolism and excretion are the major determinants of serum concentration of a given dose of product.

Knowing the concentration of the unbound fraction of a compound in plasma will assist in assessing the relevance of in vitro data. Also, the plasma concentration can assist in comparing data across animal species (note that the concentration of the parent compound and/or any active metabolite is frequently used when the unbound fraction is unknown). Knowing the concentration of the unbound compound in plasma may be used as a surrogate marker for toxicity potential if a relationship has been established between the concentration and toxicity. For example, studies evaluating barbiturate sleep time illustrate a similar effect for a given plasma concentration across animal species; barbiturate sleep times may vary among species, but each species appears to awaken at approximately the same barbiturate plasma concentration (Gillette, 1976).

When judging whether the concentration will reach levels of concern in humans in the absence of information relating dose to systemic concentration, conservative assumptions should be used. In the absence of specific data about an ingredient in humans, one should assume rapid absorption and 100 percent bioavailability and divide the dose administered by the