The Bush Administration withdrew a strict new standard for acceptable levels of arsenic in drinking water on 20 March. This raised eyebrows among some scientists, because a 1999 National Research Council (NRC) report endorsed lowering the level; recently, another study also raised warnings about arsenic. But experts say the problem is tough, because detecting risks from very small amounts of toxins is fraught with uncertainty.
Low levels of arsenic can cause bladder and other cancers when ingested over many years. To cut down the risk, the Clinton Administration issued a rule to reduce the acceptable level of arsenic in drinking water from 50 parts per billion (ppb) to 10 ppb. Christine Todd Whitman, the new Environmental Protection Agency (EPA) administrator, agrees that the standard should be reduced to below 50 ppb, as the NRC panel recommended. But she wants to review the cost of the 10 ppb standard--which could be quite high for some Western communities--and ensure that it's "supported by the best available science."
But that's a question science can't easily answer. Until recently it was impossible to set risks based on animal studies. Scientists have had to rely mainly on studies of cancer in Taiwanese villagers exposed to arsenic from wells from the 1920s to 1960s. But those arsenic levels were relatively high--200 ppb or more. To predict risks at levels below 50 ppb, researchers have had to extrapolate the data assuming a linear relationship. But if exposure to arsenic is safe below some threshold, the technique could overestimate the risk. "The lower you go, the greater the uncertainty is," says Robert Goyer, a pathologist who chaired the NRC panel.
As the EPA takes another look, one new study may support the 10 ppb standard: In the 1 March issue of the American Journal of Epidemiology, a Taiwanese research team examined cancer cases in villagers who had arsenic levels as low as 10 to 50 ppb. The study, the first of its kind, found cancer risk rose with arsenic levels even at these low exposures. "On the face of it, I think [the new study] might be quite important," says Kenneth Brown, a statistician and consultant in Chapel Hill, North Carolina.