Re: Linear-no threshold question

["John Moulder" <jmoulder@post.its.mcw.edu>]

> It has now been over a year since Cohen showed the linear-no 
> threshold theory (LNT) to be invalid... [Health Phys 68, 2:157-174, 1995] 

Sorry, science does not work that way.  Rarely does a single study ever refute 
a theory.  Furthermore, while the Cohen paper is interesting, it hardly 
"refutes" linear no-threshold.  The various limitations of that paper have 
been discussed here previously.

> Cohen's observations and analysis have now been substantiated by... 
> [Lubin et al Health Physics 69:494-500,1995].

Showing, perhaps, that there is a threshold for lung cancer induced by 
continuous low-level exposure to radon.

The questions of more general interest are whether the linear no-threshold 
model is a reasonable model for continuous low dose exposure (public 
exposure); to low dose fractionated exposure at high dose rates (most 
occupational radiation protection); or to single fraction high dose rate 
exposures (most medical diagnostic exposure).

The problem with resolving the issue for the above exposures, is that the 
cancer rates predicted by application of the linear no-threshold model to the 
Hiroshima-Nagasaki data are below the level that can be detected by the 
largest imaginable epidemiological or animal study,

> Bond, V.P., et al. in [Health Phy 68:786-792]... show 
> that there is no theoretical basis to support the LNT. 

No, what Bond et al show is that there are theoretical reasons why there may 
be a threshold for certain types of radiation under certain exposure 
scenarios.

> Health Physics is a peer reviewed journal so it must be concluded 
> that none of these papers had errors in fact or in logic, and, there 
> have been no subsequent papers by any author refuting them. 

The above could only be said by someone who has never been on a journal 
editorial board and who has not reviewed very many papers for peer-reviewed 
journals.

BTW:  For the general case of protracted exposure to low-LET radiation, it is 
relatively easy to show that linear no threshold model, as applied in BEIR V, 
predicts that no statistically significant effects will be seen in human 
populations below a total dose of around 1000 mSv.  With animal studies you 
might be able to push that down to 250 mSv.  Therefore, if a threshold existed 
at, for example 100 mSv, it would be undetectable.  This indicates that the 
linear no threshold model is inherently unverifiable over the dose range of 
interest for radiation protection.


John Moulder (jmoulder@its.mcw.edu)
Radiation Biology Group
Medical College of Wisconsin