[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Linear Hypothesis



Risk models are used for at least three purposes:

1. Establishing protection standards

2. Litigation and establishing probable cause

3. Extrapolating dose to probability of harm

Are these all the same thing?  I would say no.

1.  Establishing radiation protection standards.  The linear model is
a conservative and adequate model for setting safety standards.
Any model used for this purpose should be conservative.

2.  Litigation and establishing probable cause.  All of the models
will be argued and if the lawyers do their job the court will be
presented with a good picture of whether or not the "natural" cancer
rate is far more likely to be the probable cause than the "unnatural"
radiation exposure.  The linear model is probably appropriate to
use in these cases since it is conservative and as long as the
improbablity of low doses causing cancer is well argued it
should not lead to bad results. (Maybe I'm a little optimistic here :)

3. Extrapolating dose to probability of harm.  Estimating the number
of excess cancers in an exposed population or even the expected
probability of harm to an exposed individual.  Here is where I
believe we should only be allowed to use scientifically _proven_
data.  I believe it is inappropriate to use the linear model, at low dose,
in this instance since it has not been demonstrated to be any more
valid than the threshold or even the hormeisis models.

Perhaps if we can establish the realm of applicability
for risk modeling we can reach some consensus within the society.
The fact that there is a rift between the two sides of the 
"experts" at present makes it unlikely for regulators, government,
and the public to relax their fear of the "R" word.

In a talk to our local chapter on friday Dick Vetter said that
we shouldn't put ourselves in the position of voting on the science.
I firmly agree with that.  The question we need to address is
when are we using the linear model scientifically and when are
we using it operationally?

Dale E. Boyce
dale@radpro.uchicago.edu