Re: LNT models

[Christoph_Hofmeyr/CNS1@cns.co.za]

Radsafers,
This is a resend due to server problems during the past week and therefore
somewhat late.
Some of the correspondents, notably Field and Lubin, seem to attach a
special meaning to BEIR no-threshold  linearity, which I am at a loss to
follow, having been unsuccessful in finding a Beir here at this latitude.
In my book linear is linear and no-threshold runs through the origin of
effect versus dose, and the only variable to be determined is the gradient.
So please alleviate my blissful ignorance.  If some other form of monotonic
function is meant by LNT or BEIR LNT, so please enlighten me.  Before being
put straight, I would like to speculate a bit on a simple, mathematical
LNT, and whether it could be tested by 'ecologic' data.
The simple LNT assumption (even the ICRP admitted that it is just that) has
certain consequences, namely to give substance to the concept of collective
dose.  (Paraphrasing Prof Raabe: 'Under' the LNT assumption ...) the
stochastic consequences must then be in proportion to the collective dose.
By applying some kind of linear logic, one can argue that specific
stochastic consequences to the lung in the form of lung cancer must be in
proportion to a collective dose to lung tissue or to a surrogate quantity
which is proportional to the latter.  Since it is established that radon
and progeny deliver the major part of the radiation insult to the lung, it
would seem Dr Cohen's 'ecologic' data should suffice to compare regional
sub-units, e.g. counties, on the basis of specific consequence per
collective dose (or surrogate).  The latter can presumably be evaluated
reasonably consistently.  If this ratio is fairly constant, it would be a
vindication of the LNT assumption; if not, there will be a concern about
possible confounding factors.
I would be tempted to treat smoking (the No1 suspect) also according to an
LNT assumption and collective exposure, namely proportional to gross
cigarette tobacco consumption, since radon daughters on the tobacco leaf
are inhaled even in passive smoking.  Incidentally, how does cigarette
consumption on its own correlate 'ecologically' with lung cancer?  (If it
also bucks conventional expectations, could it be a fairly convincing
argument against 'ecological' studies?).  It must then be seen whether the
adjustment of a weighting factor between radon and smoking can equalise
consequences per collective 'dose' (radon plus tobacco) between regional
sub-units, e.g. counties.  If tobacco smoke is a much more potent
carcinogen than radon progeny, it should show up in this type of
comparison.
When comparing the two supposed main causes, tobacco consumption is
probably much more variable in time, and one should probably look at
consumption for the past 3 or 4 decades in relation to consequences.  Radon
can be assumed to be a constant factor, as it is related to uranium in the
ground, i.e. the local geology.  The extent of population migration should
be assessed as a possible confounding factor, i.e. counties with high
in/out migration rates might not be suitable subjects for a study which
involves chronic factors and a very significant average delay between cause
and effect.
Disclaimer: own musings,
Chris Hofmeyr
chofmeyr@cns.co.za


************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html