Threshold (was RE: Healthy Worker Effect vs. Hormesis)

["ALDEN N TSCHAECHE" <XAT@inel.gov>]

It is certainly reasonable to conclude unobservable effects don't exist.
However, those who espouse the LNTH are not reasonable in that respect.
They fear (emotion) that effects will be found some day and then they'd
feel awful if they had let that happen because of their not insisting
that their hypothesis was right. I have long advocated the idea of: "If
you can't observe something, it doesn't exist." But, rationality
sometimes takes a back seat to emotion (or to ulterior motives).Al Tschaeche
xat@inel.gov

*** Reply to note of 03/25/96 10:44

From: Albert Lee Vest
To: RADSAFE --INELMAIL RADSAFE

Subject: Threshold (was RE: Healthy Worker Effect vs. Hormesis)
Joe Shonka (sra@crl.com) says

>>Al (Tschaeche),
>I can't believe you.  But, you've got the controversy stirring.  In
>order to prove a statistical point, the power of the study must be
>defined.  In many cases, the statement has been made that it would
>take a controlled experiment with more than the world's population to
>establish the "linear with no threshold" theory with adequate power
>at occupational dose levels.  The converse is also certainly true.

Something I've wondered about.  Even by the largest possible (finite) study,
even assuming the single-parameter "LNT" dose-effect model, it can be shown
that health effects below a certain (finite) exposure level cannot be
detected.  Is that level then not in effect a threshold?

The principle appealed to is this:  For anything to be said to exist, one
must be able, in principle, to detect it.  Since effects below that
calculated level (whatever it is) can never be found, then it is reasonable
to conclude they don't exist.

Comments?  Is the principle above generally accepted?

Albert Lee Vest           The Ohio State University
Health Physicist    Room 103 1314 Kinnear Road Bldg
(614)292-1284                     1314 Kinnear Road
avest@magnus.acs.ohio-state.edu   Columbus OH 43212
My employer did not review or approve this message.