STATISTICAL CONSIDERATIONS IN DETERMINING ENVIRONMENTAL SAFETY STANDARDS FROM EPIDEMIOLOGICAL DATA

This paper develops the notion that environmental safety standards for human populations should be based on epidemiological data, and reviews some powerful statistical procedures which have been developed recently for this purpose. It then considers some of the limitations of epidemiological data and the ability of these new techniques to overcome them. To the extent that these limitations are inherent in the data and unmeasurable, conservative “safety factors” must be applied to the recommendations arrived at by any technique, but this uncertainty is no justification for using less than the best available analytical techniques.