Re: Deaths from fossil fuel burning air pollution

[Bernard L Cohen <blc+@pitt.edu>]

On Mon, 20 Nov 2000, Otto G. Raabe wrote:

> At 11:00 AM 11/20/00 -0600, Prof. Coehn wrote:
> >	Incidently, there is no "linear-no threshold" assumption stated,
> >or even implied. There are lots of data on exposure vs health effects, and
> >these are analyzed with regression analysis involving other potential
> >confounding factors, i.e. multiple regression. Regression analysis by
> >definition gives a linear fit to the data, but it is not "linear-no
> >threshold"; the (0,0) point is not included in the regression analysis.
> **********************************************************************
> November 20, 2000
> Davis, CA
>  This is expressed by published statements
> such as: "A 2% increase in mortality occurs for every 50 micrograms per
> cubic meter increase in particulate air pollution." This is based on
> comparison of higher PM levels with lower levels with respect to observed
> mortality. In this model, any increase in concentration, NO MATTER HOW
> SMALL can be related to an increase in mortality. A 1 microgram per cubic
> meter increase in particulate matter concentration then yields a calculated
> 0.04% increase in deaths.

	--This is meant to apply only in the region of the data used to
derive it, not to exposures "no matter how small"

 That works out to be 4 additional deaths for
> every 1000 deaths that may occur without this increase in particulate
> matter; large cities can be said to have many deaths associated with even
> this small increase in airborne particulates matter. 
> 
	--That is what the data shows, and I see no reason not to believe
it. Of course the data are not perfect and there are error bars.

> This type of calculation does not depend on the absolute value of the
> baseline value. The "zero" level is taken as no increase in concentration
> which is associated with no increase in deaths. Any non-zero increase is
> taken to yield a linear increase in risk. There is no (0,0) absolute value
> because there is always some particulate matter in ambient air and there is
> no zero death rate at any level. However, it is implied by this calculation
> that every microgram per cubic meter of particulate matter causes a linear
> increase in death rates with no threshold. I call that a Linear
> Non-Threshold (LNT) risk model.

	--You can call it what you like as long as you don't apply it
outside the region where there are data. Names don't matter, but LNT has a
bad name because it is applied outside of the region where there are data.
That aspect of the bad name for LNT does not apply to the air pollution
studies.
	By applying this bad name to these studies, you are essentially
saying that all regression analysis is tainted by that bad name. You
should know that a great deal of economic analysis is done by regression
analysis, so the economic well being of our country is dependent on it.
That is just one example - regression analysis is done in many areas,
including most epidemiology studies, psychology, social work, etc. Are all
these to be tainted by the bad name of LNT derived from utilizing it to
make predictions for health effects of radiation far outside the region
where there are data? 

Bernard L. Cohen
Physics Dept.
University of Pittsburgh
Pittsburgh, PA 15260
Tel: (412)624-9245
Fax: (412)624-9163
e-mail: blc+@pitt.edu


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