Public health standards often rely on animal data, despite the availability of more relevant human data. The recent case of health recommendations regarding safe levels of perchlorate in drinking water is used as a case study illustrating how existing human data can offer a reliable basis for inferences about effects on human health. The direct effect of perchlorate exposure is the inhibition of iodine uptake by the thyroid gland. Subsequent effects on thyroid hormone production potentially may lead to adverse neurological and cognitive development or thyroidal effects. Existing human data are sufficient to determine a level of perchlorate exposure that will protect against these effects. The published literature on humans and perchlorate exposure includes medical, occupational, epidemiological, ecological, and laboratory studies. A systematic review of that literature provides a solid, robust, and consistent assessment of human risk from perchlorate exposure. This yields, in regulatory or risk assessment terminology, exposure levels for a lowest observed adverse effect level (LOAEL), a lowest observed effect level (LOEL), and a no observed effect level (NOEL). In vitro studies of the cellular mechanism support the human findings. The therapeutic level for perchlorate is in the range of 100–1000 mg/d. At occupational exposures of up to 35–50 mg/d, no effect on thyroid hormones is seen. Iodine inhibition has been demonstrated at levels down to 1.4 mg/d. No effect has been observed in persons exposed at 0.5 mg/d or below. The data indicate a threshold of about 0.4 mg/d or 200 μg/L if in drinking water. Ecological and epidemiological studies of long-term perchlorate exposure at levels up to 100 μg/L find no effects. Rat data should not trump the totality of human data to assess human risks. More than 50 years of clinical, occupational, and epidemiological experience with the effects of perchlorate exposure to humans is available and sufficient to identify safe exposure levels. These human data provide the most relevant basis for establishing public health standards and obviate the need to extrapolate from animal models that may not accurately represent humans.
Read full abstract