statements that are not possible to document. Use of primary literature also increases credibility of resources when persons or organizations involved in producing them might be perceived to have a bias or conflict of interest.
Does the review consider all types of information available, including data from in vitro studies and animal studies, as well as information about the safety of closely related plants or substances? A review that attempts to integrate a variety of data types, especially in the absence of good quality human data, will be a more appropriate resource for assessing safety. Because it is scientifically appropriate to consider concerns raised by animal data, in vitro studies, or information about related substances (see Chapters 4–7), reviews that conclude a substance is safe by focusing exclusively on human data should not be considered as adequate sources of safety conclusions. The converse is not true. That is, reviews that conclude a risk exists by exclusively focusing on human data are acceptable.
Has appropriate scientific expertise and objectivity been used in weighing different types of information? Good reviews will explain why some information is considered more important than other information in reaching overall conclusions. Clearly stating the logic that underpins the data interpretation enables other experts to understand the basis for the evaluative judgment and determine whether appropriate objectivity and scientific expertise were used.
What are the limitations to the safety review? Does the review explain why it may be difficult to make a conclusion about the safety of an ingredient? Does the review describe where insufficient, inconsistent, or inadequate data preclude an accurate assessment of safety? Discussion of limitations in data, such as how much is known and how definitively it is known, is useful in understanding safety. Also, an appropriate discussion about the limitations in interpreting the available data lends credibility to the review. Limitations may be of two types. First is the individual study’s limitations, such as limitations in interpreting the data or in experimental design, sometimes described by the study authors themselves. Another type is derived from developing the review itself, such as difficulty in interpreting data from foreign language sources or inconsistent data from different sources.
Were the strategies used to search the literature adequately described? A good review will describe how the search for pertinent data was performed. Ideally, this description will include a list of databases searched, when they were searched, what search terms were used, and if there was a strategy for selecting information to review. This allows the user of the information to determine if the breadth and depth of the search was appropriate.
How current is the information? A good review includes relevant current information, in addition to older information published in repu-