Estimation of toxic hazard—A decision tree approach

Safety evaluation is caught in a frustrating circle. It is neither possible nor sensible to try to obtain the information needed to assess every imaginable toxic risk associated with every substance, and pursuit of greater safety therefore demands the setting of priorities as well as sensible limits for investigation. To do this with confidence requires possessing the very information that is lacking and that can be won only slowly on a few substances at a time, with significant uncertainty and at considerable cost. This requires priorities, and completes the circle of frustration. Individual texicologists deal with this problem by using ‘experience’, a personal synthesis of accumulated knowledge of structure-activity relationships, metabolic mechanisms, chemical reactivity, human exposure and other relevant information. Such expert judgement is often very effective in distinguishing potential risks worth pursuing from problems on which effort would be wasted but, because it is usually so inexplicit and subjective, it is seldom able to invoke the public confidence most decisions now require. This paper proposes a procedure for making a significant part of this process rational, public and explicit. It uses much currently available toxicological data to validate the procedure, which consists of a ‘decision tree’ of 33 questions, each answered ‘yes’ or ‘no’. Each answer leads to another question or to final classification into one of three classes (I, II and III) reflecting a presumption of low, moderate or serious toxicity. The tree is organized into branches dealing with major chemical classifications and is intended for use with all ingested, structurally defined organic and metallo-organic substances. Answering the questions requires chemical or biochemical training, and relies primarily on features of chemical structure. Occurrence in body tissues and fluids, and natural occurrence in food are also involved. The logic of the tree rests heavily on known data on metabolism and toxicity. The classification according to presumptive toxicity can be combined with knowledge of human intake to provide for each substance a ‘protection index’, which can be used to establish priorities and to define tentatively the extent of appropriate testing. The procedure has been applied to a large number of pesticides, drugs, food additives and industrial and environmental chemicals of known biological properties. So far it has not resulted in any underestimation of toxicity, and it appears to provide a practical means for discriminating effectively among different levels of probable hazard.