Do observations of chemical hazards at high exposure levels
predict risk at much lower levels? Or are there reliably “safe”
(that is, zero-risk) doses or exposure levels for most chemicals
other than carcinogens? To lay down the gauntlet, this column
argues that the concept of a “threshold” in a population violates
the principle of the exposure-response relationship. I agree
there are population doses with minimal risk, perhaps practical
thresholds for concern. But assuming a threshold assumes
there is an exposure region where increasing exposure doesn’t
increase risk.
Our jobs as practitioners include communicating the risk
of a chemical exposure at levels well below those where a
hazard has been identified. Paracelsians fondly say that
“everything” is a poison at some dose, as a way soft-soaping
people exposed to a chemical already shown to a be a poison.
(I think Paracelsus used that argument to defend using mercury
as a medicine, based on an alchemical theory; probably he
said or wrote it in Latin.) Opining about causation where one
person among a larger group of similarly exposed individuals
becomes ill is another side of that coin. In some situations,
the illness is clearly associated with exposure to the chemical
in some setting; in others, past evidence of association may
be shaky.
As practitioners, we rely mostly on authorities like government or scientific organizations to tell us the level of exposure
where there’s a concern or probability of adverse effect. Since
OSHA and NIOSH have done little in the past decade (or more)
to bring exposure rules or recommendations in line with
emerging science, the main sources of such data are the EPA
Integrated Risk Information System (IRIS) reference doses and
concentrations ( www.epa.gov/iris/) and the Agency for Toxic
Substances and Disease Registry (ATSDR) minimal risk levels
(MRLs) for hazardous substances ( www.atsdr.cdc.gov/mrls/
index.html).
I tell my students to go to IRIS and ATSDR whenever they
have an identified chemical exposure level to evaluate. IRIS
calculates a unit cancer risk for carcinogens, based on linear
low-dose extrapolation, and generates reference doses and concentrations for non-cancer endpoints by applying uncertainty
factors to a no-observed adverse effect level (NOAEL) as a point
of departure. Some of these values are aggressive ( 5 mg/m3 for
toluene), and some are very aggressive (0.1 mg/m3 for xylenes).
This column provides arguments for applying these aggressive
limits in the occupational setting.
Defining “Zero-Risk”
We are taught that all chemicals exhibit an exposure-response
relationship: the lower the dose, the lower the risk. Is there a
threshold—that is, the upper bound of zero-risk doses? A
threshold dose implies a dose region in which there is no exposure-response relationship—within this region, increasing the
dose carries no increased risk because the risk is zero. This region is a biological black hole: dose goes in, but no response
comes out.
What evidence is there for a threshold? Proving the threshold (in a population exposure-response relationship) is as difficult as proving any negative. I teach the concept of “Limit of
Direct Observation” (LODO)—the power of various methods to
see the toxic potential of an exposure, if it were there. The
LODO for laboratory studies is a risk of about 1 in 10 against a
zero background, the risk at the NOAEL or the benchmark dose.
A background risk in control animals—for example, liver cancer
in mice—moves the LODO upward. Some special designs (
discussed below) might do a little better. (Remember that the
Supreme Court-derived border of “significant risk” is 1 in 1,000,
or 0.1 percent.)
