Re: Severe limitations of ecologic data

[BERNARD, L, COHEN]

	I agree that our discussions on RADSAFE have reached an impass. I

will not offer further comments unless they are in response to a concrete

hypothetical example of what can be wrong with my procedures, or in

response to a question about material in my papers.



On Tue, 1 Jul 2003 epirad@mchsi.com wrote:

>

> I have made every effort to explain my concerns in detail (including many

> references) regarding your use of ecologic studies to test the LNT. In my

> opinion, these discussions have little to do with the LNT, but rather the

> limitations of using ecologic data. For example, how can surrogate county

> smoking data be used to treat the confounding in your relationship for people

> who have smoked 30 years or more at various rates per year? Your surrogate

> data represents a relatively short time period and we have no idea whether or

> not it actually reflects the relevant smoking information in a county.  What

> evidence can you provide to me that your ecologic data is accurate enough to

> test the LNT?  Your previous statements that your large data set allows the

> errors to average out is scientifically ungrounded.

>

> I have presented evidence that your smoking data does a poor job of explaining

> the variation in lung cancer noted for your counties. Dr. Puskin has presented

> further information that suggests your findings are due to residual

> confounding by smoking.  Further, I have presented information that your

> mortality estimates do a poor job of predicting the actual incidence rates for

> a county for the time period of interest.

>

> Your continued comparisons to case-control studies have little to do with

> determining whether or not your data are rigorous enough to use to test the

> LNT. Do you understand that ecologic studies are subject to cross-level bias

> which can cause unbounded bias in either the negative or positive direction,

> while this is not a problem with case-control studies? You already stated you

> can not use your ecologic study to examine the dose response for radon.  As

> Dr. Mossman pointed out in this months HP Newsletter, if you can not use it to

> examine a dose response for radon you can not use it to test the LNT since

> what is the LNT but a dose response.

>

> Perhaps your lack of understanding of my views is a result of a general

> failure to acknowledge the limitations of your ecologic data. If you can not

> acknowledge the limitations of ecologic data, there really is no need to

> continue this dialogue.  Let's both move on to more constructive use of our

> time.

>

> Respectfully,  Bill Field

>



************************************************************************

You are currently subscribed to the Radsafe mailing list. To unsubscribe,

send an e-mail to Majordomo@list.vanderbilt.edu  Put the text "unsubscribe

radsafe" (no quote marks) in the body of the e-mail, with no subject line.

You can view the Radsafe archives at http://www.vanderbilt.edu/radsafe/