Re: LNT and Public Fear

[John Goldsmith <gjohn@bgumail.bgu.ac.il>]

Dear Bob:
	Your conclusions, 1 and 2 are both correct, but the reasons are a 
bit different.  As I understand the scientific method in general, it     
cannot prove a negative, but can only disprove a null hypothesis.  
Because epidemiological findings can only refer to a specific population, 
even disproving a null hypothesis cannot be treated as a generalization, 
which is what a threshold would imply.  Epidemiology can fail to disprove 
a null hypothesis, say that there is no more leukemia among residents near 
a radiation source than would be expected on the bais of general 
population observations---a "negative"result , or it can disprove 
it, a result consistent with leukemia excess. But these results are only 
valid for the population and exposure history specified.  From a 
substanital number of such studies, a generalization can be hypothesized.
Among epidemiologists, there remain variations in the criteria by which 
an association is to be interpreted as causal, and the acceptable levels  of 
type I and Type II error.  However, there is widespread agreement that
it is not possible for epidemiological studies to define  thresholds for 
carcinogenic agents.
   gjohn@bgumail.bgu.ac.il    John Goldsmith, M.D., M.P.H.
                          Professor of Epidemiology

On Mon, 14 Oct 1996, Bob Hearn wrote:

> Dear John-
> 
> If I understand the thrust of your comments, the position is 1)that the
> tools of epidemiology cannot be applied to prove a negative finding or
> absence of phenomenon; and/or 2)epidemology cannot confirm that
> _absolutely_no_ occurence will be manifested below a threshold. I believe
> that the only claim to be made is that of certainty at a given level of
> confidence, a limitation which also applies to experimental findings. I
> would think that an epidemiological result can be obtained for lack of
> occurence below a threshold as well as it can for positive observations at a
> higher level, subject to the same complications of population variability.
> Please clarify or indicate if you are refering to practical limitations of
> statistics at baseline levels.
> 
> Regards,
> 
> Bob Hearn
> 
> At 02:28 AM 10/14/96 -0500, you wrote:
> >Dear Mark:
> >     While I agree with your conclusion and much of your logic, I must,
> >as an epidemiologist, deny that epidemiological observations are the 
> >sources of thresholds for hazardous chemicals or carcinogens.  It is 
> >experimental work that allows, under the circumstances of the experiment,
> >the statement that a threshold has been established.  
> 
> >                                                 Population variability
> >in responsiveness is the major logical reason for epidemiology being 
> >unable to establish a threshold.  Epidemiologic observations often can 
> >affirm that in a comparable population to the one studied, effects will occur 
> >at a given level of exposure, but it can     never     show that no
> unfavorable 
> >effects will occur below a given exposure level.
> 
> > (Which is what I 
> >understand a threshold to mean).  Finally, epidemiology can, under some 
> >circumstances, estimate the dose or exposure which would be associated 
> >with a lifetime increased incidence of a small number of cases, say 2-4 
> >per million.  In my view, at least, thresholds can only be based on 
> >epxperimental exposures, and are valid for sufficiently similar exposures;
> >it follows that experimental exposure of human populations to radiation 
> >are fraught with ethical problems, hence, thresholds for such exposure 
> >are not feasible.
> >    gjohn@bgumail.bgu.ac.il   John Goldsmith, M.D., Professor of Epidemilogy,
> >                              Ben Gurion University of the Negev
> >On Sun, 13 Oct 1996 MarkH@MAIL.KIDS.USYD.EDU.AU wrote:
> >
> >> Friends, 
> >> 
> >> I feel that Dave Scherer's thoughts are on the right track.   
> >> 
> >> As in the case of hazardous chemicals and/or carcinogens, a "threshold"
> >> will be determined from epidemiological observations (i.e. of populations).
> >> The risk threshold therefore applies to the population, and cannot directly
> >> be applied to the individual. How many members of the public will
> >> appreciate this subtlety ? 
> >> 
> >> I suspect that, were a threshold introduced, many members of the public
> >> would still reason along the lines : "if 20 units is not safe, then can
> 10 units
> >> be completely safe ?".  (I guess this might not be the case if there were
> >> orders of magnitude of difference between the threshold and the typical
> >> exposure.) 
> >> 
> >> The real solution lies in educating the public.
> >> 
> >> 
> >> Mark Hanlon,
> >> Radiation Safety Officer,
> >> The New Children's Hospital,
> >> P.O. Box 3515,
> >> PARRAMATTA, NSW, 2124
> >> Australia
> >> 
> >> 
> >> 
> >
> >
>