Essentially, the debate is about keeping as broad a safety margin as possible.
If it were trivially-cheap to analyze water for the presence of lead--let's say it cost 1 penny per hundred billion gallons of treated water to remove and verify lead content down to the 1/1,000,000 ppb level (that means any given lake-sized volume of treated water has a high likelihood of having zero lead atoms in it period)--we would mandate that. Why wouldn't you?
What failures in measurement expose us to additional radiation? What procedures (e.g. radiology) do we go through that exposes us to additional radiation? For a population of hundreds of million, is this level of radiation prone to cause a hundred more incidences of cancer (trivial) on its own, before interacting with other factors?
One person in America dying every year might be a triviality. If it costs millions of dollars to prevent that, well, let's not do it: you'll save more lives investing that in charity and anti-poverty measures. If it costs pennies per year, then yes let's do that.
"Pennies" quickly becomes "dollars" and "millions of dollars" as you add zeroes onto the end of that one person. 1,000 persons per year? Maybe we want to invest several million dollars into this--especially since "dying" isn't binary when you get past bullets to the head. Even highway safety measures come down to death, dismemberment, or property damage.
It's a matter of risk--a highly-technical concept nobody seems to know all that much about.