Cohen's questions

[epirad@mchsi.com]

Dr. Cohen,



If you truly think the article below is outside your area of expertise and do 

not understand the significance of the study, performed at Johns Hopkin's, to 

the quality of your mortality data you use (as compared to SEER incidence 

data); please consider discussing both the significance of the article below 

and the limitations of your data with an epidemiologist you trust. 



Bill Field

---------------------------------------------------



Dr. Cohen stated, My lung cancer data are from National Center for Health  

Statistics. If those data are no good, why are they collected? 



They were never intended for use to test the LNT or to use to "treat" 

confounding. However, my point was that the SEER data is more reliable and it 

is in very poor agreement with the data you use.  We showed this was clearly 

the case with the Iowa data you have never been able to explain. 

------------------------------------

Consider this finding:



Arch Intern Med. 2001 Jan 22;161(2):277-84.    

  

Problems with proper completion and accuracy of the cause-of-death statement.



Smith Sehdev AE, Hutchins GM.



Department of Pathology, The Johns Hopkins Hospital, 600 N Wolfe St, 

Baltimore, 

MD 21287-6901, USA.



BACKGROUND: Mortality statistics are largely based on death certificates, so 

it is important that the data on the death certificate is accurate. At our 

institution, clinicians complete cause-of-death statements (CODs) prior to 

autopsy. Since May 1995, separate CODs have been included in autopsy face 

sheets. 



METHODS: Clinical and autopsy-based CODs filled out separately on 494 

cases between June 1995 and February 1997 were compared for proper reporting 

and accuracy using the published guidelines and definitions of immediate, 

intermediate, and underlying causes of death put forth by the College of 

American Pathologists and the National Center for Health Statistics. 

RESULTS: Of the 494 death certificates, 204 (41%) contained improperly 

completed CODs. Of these, 49 (24%) contained major discrepancies between 

clinicians' and pathologists' CODs. Of the 494 death certificates, 290 (59%) 

had properly completed CODs. Of the 290 properly completed CODs, 141 (49%) 

contained disagreements: 73 (52%) on underlying CODs; 44 (31%) on immediate 

CODs; and 47 (33%) on other significant conditions (part II). 



CONCLUSIONS: The reliability and accuracy of CODs remain a significant 

problem. Despite its limitations, the autopsy remains the best standard 

against which to judge premortem diagnoses. The CODs of the death certificate 

may be improved if death certificates are completed in conjunction with the 

postmortem examination and amended when the autopsy findings show a 

discrepancy.



What is your response to this study?

-----

Dr. Cohen states that, the issue of measuring radon now to explain lung cancers

presumably due to radon exposures many years ago is present also in 

essentially 

all case-control studies. 



Please understand that non differential radon exposure misclassification in a 

case-control study lowers the risk estimates to no association, but in an 

ecologic study can produce unbounded bias in a negative or positive 

direction.  

 

Dr. Cohen states, "My very extensive studies have shown that nothing except 

urban-rural differences correlate strongly with radon urban-rural differences 

correlate strongly with radon".  



Perhaps you need to more concerned with what correlates with smoking such as 

urban/rural differences since that is driving the majority of the risk versus 

effect.  I presented a potential way for you to perform the analyses to test 

the plausibility of my suggestion (using a Monte Carlo Analyses or a 

Sensitivity Analyses), now the rest is up to you. 





Sincerely, 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/