[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Cohen's Refutation of LNT



Sorry, what I meant to ask was if anyone had formulated an individual risk

model (hypothesis) based on the US data (Cohen's or EPA ...)



Thanks for the reference below. For a specific question about that paper:



What was the hypothesis that was generated from the fact that "Within low

smoking

 counties, rates for all lung cancer and small cell carcinoma were

 significantly lower (p < 0.05) in the high radon counties relative to the

 medium and low radon counties." and how was it tested in the later Iowa

study?



An obvious hypothesis (but not the only possible one) would be: "In low

concentrations, radon can prevent the occurrence of lung cancers in

non-smokers."



To test this hypothesis with a case control study, you would limit the study

to non-smokers, because your hypothesis only deals with non-smokers. You

would also have to choose your exposure categories so that an increased risk

in "low" exposed individuals could be detected. That would probably mean

limiting your study to low radon, low smoking counties. (After all, you are

trying to figure out what is killing the people in these counties.)



The obvious problem with a study like this is the small sample size.

However, including individuals or populations (e.g. smokers) that are not

relevant to your hypothesis actually makes the statistics worse.



Science should go like this:



1.    You make an observation. (via ecological study or whatever.)

2.    You come up with a formal, preferably quantitative, hypothesis to

explain your observation.

3.    You crunch the numbers to ensure that your hypothesis could actually

explain your observation. If it can't, go back to step 2.

4.    You design an experiment (case-control study or whatever) to test your

hypothesis.

5.    You clearly state the hypothesis to be tested and how the particular

experiment will confirm or refute the hypothesis.

6.    You perform the experiment and present the results.

7.    You show how the results of the experiment either confirm or refute

the hypothesis.

8.    If your experiment refutes the hypothesis, go back to step 2.



All I can see are steps 1 and 6.



Regards,

Kai



----- Original Message -----

From: "field" <bill-field@uiowa.edu>

To: "Kai Kaletsch" <info@eic.nu>; "Rad health" <healthrad@HOTMAIL.COM>

Cc: <radsafe@list.vanderbilt.edu>

Sent: Sunday, February 10, 2002 12:10 PM

Subject: Re: Cohen's Refutation of LNT





> Kai,

>

> We published an ecologic study very early on before we followed up with a

> case control study.

>

> : Health Phys 1994 Mar;66(3):263-9 Related Articles, Books, LinkOut

>

>

> Residential radon exposure and lung cancer: evidence of an urban factor in

> Iowa.

>

> Neuberger JS, Lynch CF, Kross BC, Field RW, Woolson RF.

>

> Department of Preventive Medicine, University of Kansas School of

Medicine,

> Kansas City 66160-7313.

>

> An ecological study of lung cancer, cigarette smoking, and radon exposure

> was conducted in 20 Iowa counties. County-based lung cancer incidence data

> for white female residents of Iowa were stratified according to radon

level

> and smoking status. Cancer incidence data for the period 1973-1990 were

> obtained from the State Health Registry of Iowa. Smoking level was

> determined from a randomly mailed survey. Radon level was determined

> according to an EPA supported charcoal canister survey. Within low smoking

> counties, rates for all lung cancer and small cell carcinoma were

> significantly lower (p < 0.05) in the high radon counties relative to the

> medium and low radon counties. However, within high smoking counties,

rates

> for all lung cancer, adenocarcinoma, and small cell carcinoma were

> significantly higher (p < 0.05) in the high radon counties relative to the

> low radon counties. Variations in socioeconomic data for these counties,

> available through the 1980 and 1990 census, did not explain these results.

> Lung cancer rates also were significantly increased in urban counties even

> after holding smoking status constant. Multivariate analyses revealed

> significant interactions between smoking, urbanization, radon levels, and

> lung cancer. The results of this hypothesis generating study will be

tested

> in a case/control study now ongoing in Iowa. Analysis will need to include

> separate evaluations by smoking status, radon level, and residence in

urban

> or rural areas for the major morphologic types of lung cancer.

>

> Regards, Bill

>

> ----- Original Message -----

> From: Kai Kaletsch <info@eic.nu>

> To: Rad health <healthrad@HOTMAIL.COM>

> Cc: <radsafe@list.vanderbilt.edu>

> Sent: Sunday, February 10, 2002 10:51 AM

> Subject: Re: Cohen's Refutation of LNT

>

>

> > > He would say the ecologic study was good to form hypotheses, but the

> > > hypotheses have been proven wrong by the more rigorous case-control

> > studies.

> >

> > Could anyone please provide a reference where a hypothesis has ever been

> > formed based on the ecological radon data? So far I have only seen

> > regressions of the COUNTY risk. I am not aware of a single INDIVIDUAL

risk

> > model that has been constructed to explain the county data.

> >

> > Could anyone please provide a reference where such a hypothesis has been

> > tested by a case control study. The studys I have seen simply find risk

> > coefficients for a linear model. Would anyone really hypothesise a

linear

> > individual risk model by looking at the county data?

> >

> > Thanks,

> >

> > Kai

> >

> >

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

> > 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/

>



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

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/