Re: Mr. Ford and Iowa Radon Lung Cancer Study

[Michael Ford <michaelford@cox-internet.com>]

<Note:  We'll now dispense with all titles>



Bill, you're right.  You're absolutely right.



Is that better?



You're right, I'm stupid.  I'm thick-headed.  I'm dumb.  I'm slow.



I need, not just a few, but several semesters (perhaps years?) of statistics

and epidemiology to be able to stand in your shadow.



"You da man."



I guess I'm blind too.  Blind because I have yet to see any of the treatment

you describe below also described in your study (i.e., intervals).  Or I

must not be clairvoyant enough to read between the lines to see what you

obviously intended.   Since you mentioned your statistician, did he sign off

on the level of precision you've used in your numbers?  I'd love to see the

paper where you handle your propagation of error for WLM exposures.  Again,

given the level of uncertainty you have to deal with, especially occupancy

factors, how in the world do you justify reporting to two decimal places?

(16.95 WLM)  



Do you maintain that grouping does not affect outcome?  Wouldn't an even

grouping put 20% in each exposure category?  Instead you've chosen 15%, and

4 @ 21.25%?  Are those uniform intervals or "non-uniform intervals?"



Am I just being stupid again?



By the way, I am pleased to see that you also feel that the EPA's 4 pCi/L

threshold is "arbitrary."



You broach the topic of statistical power.  Given this is a

population-based, case-control study, what is the power of your study with

the relatively small number of cases and controls?  I noticed you did not

reply to my question on statistical significance that you have maintained is

an erroneous allegation.



On the smoking business, Bill, you're probably going to have to go to the

"blue-in-the-face" level (ref: Field: "I really do not know how many times I

have to say this to be understandable to you.")  Your study is identified as

a population-based, case-control study.   I have NEVER maintained that you

MUST COMPLETELY match for smoking.  However, I do maintain that you should

have at least ATTEMPTED  to match for smoking.  Your data shows that you did

not.  In fact, the cases and controls have either non-linear, inverse

(Smoking) or markedly different (Years since cessation of smoking) patterns

of smoking behavior.



I have discussed the latter in a previous post when showing that other

studies have at least ATTEMPTED to match for smoking.  For the uninitiated,

the percentage ratios of %cases:%controls from the IRLCS fall out as below:



Smoking (pack years) [from Table 1, page 1096, Field et al]



                        %Cases       %Controls   %cases:%controls

0-19                13.2                    49.5                    27%

20-39               37.2                   30.0                    124%

40-59               28.6                   28.6                    173%

>= 60               21.0                     4.0                    525%



Modeling is great, wonderful, and fine..... if you actually believe that the

results of the model depict some version of the truth.  I guess I'm just too

stupid to believe in the ability of a model to control for confounding when

we don't actually know 1) what  the mechanism is for lung cancer induced by

smoking, and 2) IF THERE IS A MECHANISM for lung cancer induced by radon at

these exposure levels  AND when the case-control "Smoking" data is INVERSELY

related!



Especially given the relatively small numbers used in this study, you have

to attempt to match.  Get more controls!



Yes, Bill.  You are correct.  I am stupid, but perhaps not SO stupid that I

don't see through the BS (Blowing Smoke) of the "superior knowledge"

argument.  If you've been lulled into actually believing that your modeling

mimics reality to the point that you can deduce a radon and lung cancer

relationship in the background of your cases' overwhelming smoking patterns

... well, we're just not going to find any common ground here and you're

probably going to keep thinking that I'm an idiot.  I'll try not to lose any

sleep.



As for your pleasant entreaties,  I would gladly obliged, but, Bill .....

you kinda hurt my feelings, so I just "teared up" and started to type.



Yes, Bill.  I would love to end this discussion and move to the Billet

forum.  But it's a bit disingenuous on your part to label me as an idiot and

not expect a reply.  I haven't stooped to those tactics and I didn't expect

it from you. 



As for the letter you have requested, your Billet letter singled out Dr.

Klein (not Mr. Klein) and now you have singled me out in this forum (your 24

JAN 02 post).  As a result, I see no reason for the entire TRAB to be

dragged into the quagmire of insult that has ensued, and you really have no

room to make that a condition of the debate.



So if you can manage to keep the personalities out of it and stick to the

facts of the debate, we can end this fairly quickly and I can get to writing

the letter!



Have a great Day/evening/weekend,

v/r

Michael

Idiot-at-large

4th Grade Graduate

Mouth Breather





on 1/25/02 3:20 PM, Field, R. William at bill-field@UIOWA.EDU wrote:

> Michael,

> 

> I keep answering these questions, but you do not appear to understand the

> answers.

> 

> Let's try your first question once again.  How did we decide to place 15% of

> subjects (cases and controls made up this 15%, but not in equal number) in the

> upper exposure categories.  Well after the data went through QA checks, we

> archived the data.  I then sat down with the statistician and we discussed

> exposure categories.  I told him I did not want to make artificial exposure

> categories (for example > 4 pCi/L) as some studies had done.  I asked him what

> percentage of subjects (either cases or controls) would need to be in the

> upper exposure category for adequate statistical power.  He thought, at least

> 15% of the subjects should be there.  Therefore, we took the top 15% of

> subjects based on their cumulative radon exposure and placed them in the upper

> exposure category.  The remaining categories were split as evenly as possible

> by cumulative exposure.  The 16.95 WLM simply represented the lowest exposure

> for the top 15%.  This was done a priori and it was the first time we actually

> saw the risk estimates by category.  The National Pooling paper will have the

> results for the Iowa Study based on different exposure categories that form a

> common format for all the N. American Studies.

> 

> I really do not know how many times I have to say this to be understandable to

> you.  When you do a population based study, it is impossible to completely

> match for smoking up front.  Do you match on years smoked, smoking rate, year

> started smoking, year ended smoking, etc. - if you match on one of these, the

> other factors are not matched.  A large proportion of studies adjust for

> smoking, not just studies on radon.  Smoking impacts many types of cancer and

> all the studies must adjust for smoking.  This is a routine process. The more

> you match, the more you get away from a population-based study. It is not

> smoke and mirrors, but it would likely take you several semesters of

> statistics and epidemiology to understand how it is done.  In the Iowa study,

> we also performed checks for residual confounding from smoking and no residual

> confounding was noted.

> 

> I have responded to your concerns (from my perspective they were in fact gross

> misrepresentations) about the Iowa Study now in both a letter in the Billet,

> emails directly to both you and Mr. Klein, and on Radsafe.  If you want to

> carry on this discussion further, I respectfully ask that the discussion move

> off this forum and respectfully ask that you either write a letter to the

> American Journal of Epidemiology or draft a letter for the Billet that has the

> support of the Texas Radiation Advisory Board.  Surely, if you have time for

> these discussions on Radsafe, I would think you also have time to draft such a

> letter.

> 

> Regards, Bill Field

> 

> 

> ******************************************************************************

> *********

> R. William Field, M.S, Ph.D.



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

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/