Artificial Food Color Additives and Child Behavior: Weiss Responds

The Food and Drug Administration's (FDA) response to my commentary (Weiss 2012) reflects the wide gulf between how the FDA translates " weight of evidence " into regulatory policy for artificial food colors (AFCs) and how it is translated into meaning ful action on behalf of public protection. The FDA essentially took the position that for a study to be considered as evidence of adverse effects, it must be totally free of uncertainties. The study by McCann et al. (2007) played a large role is provoking the FDA review, but for that study, like almost any epidemiological study, it would be difficult to meet that absolute criterion. It is why Environmental Health Perspectives (EHP) publishes so many such studies addressing the same question (e.g., air pollution). But isn't it fair to ask whether any of the negative AFC studies meet that criterion? In their critique, the FDA faults McCann et al. (2007) because they characterized " … a treatment effect as adverse when it may, in fact, fall within the normal range of childhood behavior. " This is an issue discussed over and over again in the pages of EHP. Take the example in my commentary (Weiss 2012), modeled on numerous publications in the lead literature (e.g., Lanphear et al. 2005): If developmental exposure to low levels of lead reduces a population IQ (intelligence quotient) by 3 points (3%), from, say, 100 to 97, it is taken as evidence of a major adverse effect. Both scores, of course, fall within the normal range. The same criticism is used by the FDA to dismiss the effect size calculations; that is, the altered behavioral activity seen in published data lies " … in the range of normal activity for children. " The FDA finds the study by McCann et al. (2007) lacking because the authors relied mainly on parental observations. A high proportion of child development research, in fact, enlists parents as observers ; hundreds of validated inventories and questionnaires are based on parent ratings. They are the observers, of course, who see the most extensive samples of the child's behavior, especially with younger children. This is the reason I chose parental observations for my own food color study of young children (Weiss et al. 1980) and why we relied on parent ratings for our study of how phthalates mold play behavior in preschool children (Swan et al. 2010). It is difficult …