AW: [ RadSafe ] Nuclear mortality

Wed Aug 6 07:23:42 CDT 2008

Ross,

thank you for this important reference:

Atkinson W D, Law D V, Bromley K J.
A decline in mortality from prostate cancer in the UK Atomic Energy Authority workforce
J Radiol Prot 27#4(2007)437-445

The last sentence of their abstract merits singular recognition.

"Our conclusion is that the original observation might have been a chance finding among many significance tests." 

In the statistical literature this perennial and pervasive predicament of many radio epidemiological studies is called the "multiple testing" problem, which unfortunately has not yet found a universally reasonable solution. 

Applying the 'zero order' solution of Bonferroni by dividing type I error probabilities (e.g. 5%) by the number of tissues/organs tested would discard all (putative statistically significant) associations ever reported of cancer incidence/mortality with low dose occupational radiation exposures as statistical noise - even where one-sided tests (which already postulate a negative association) were applied and even if only an unusual low confidence level of only 90% would be stipulated. The latter two tricks, which facilitate the detection of statistically 'significant' associations, are not uncommon in radio epidemiological studies. (Honni soi qui mal y pense).

For me, their paper corroborates another personal 'bias' or long-standing suspicion. 

The know refuted earlier statistically significant associations of prostate cancer with radiation exposure came from case-control (or nested case-control) studies of the UKAEA workforce. I.e., one part of the study population is selected by the very fact of having contracted the disease under study. A frequent observation with such case-control studies is that whenever the same or a comparable population is studied in a cohort study, the significant associations found in the case-control study evaporate. Here once more Atkinson et al. report such an instance.

In principle, if properly done, both study designs should yield - within statistical uncertainty - the same results. In practice, therefore, it appears to me that - ceteris paribus - it must be much more difficult to account properly for possible pitfalls in case of case-control studies than it is in cohort studies. This may be the reason why in the "levels of evidence" of the Oxford Centre for Evidence-based Medicine results from cohort studies always rank above case-control studies. 

Based on (only) a case-control study are also the findings on prostate cancer after diagnostic radiation exposures recently advertised in radsafe (Otto Raabe, Do 31.07.2008 20:30).

P Myles P, Evans S, Lophatananon A, Dimitropoulou P, Easton D, Key T, Pocock R, Dearnaley D, Guy M, Edwards S, O'Brien L, Gehr-Swain B, Hall A, Wilkinson R, Eeles R, Muir K.
Diagnostic radiation procedures and risk of prostate cancer.
British Journal of Cancer 98(2008)1852-1856

The authors carefully stop short of claiming a causal relation behind the association they found (although they somewhat uncritically quoted from papers which do exactly this). 

Finally, to my knowledge all studies yielding a statistically significant association between lung cancer and exposure to domestic radon are case-control studies again. 

My personal bias, as I readily admit it to be, is once more intensified by Atkison et al.:

Beware of conclusions based on case-control studies only.

Regards, Rainer

Dr. Rainer Facius
German Aerospace Center
Institute of Aerospace Medicine
Linder Hoehe
51147 Koeln
GERMANY
Voice: +49 2203 601 3147 or 3150
FAX:   +49 2203 61970

-----Ursprüngliche Nachricht-----
Von: radsafe-bounces at radlab.nl [mailto:radsafe-bounces at radlab.nl] Im Auftrag von Ross Beveridge
Gesendet: Mittwoch, 6. August 2008 02:01
An: 'Steven Dapra'; radsafe at radlab.nl; 'Joel Cehn'
Betreff: RE: [ RadSafe ] Nuclear mortality

The Journal of Radiological Protection, Volume 27, Number 4, December 2007 carries a article titled 'A decline in mortality from prostrate cancer in the UK Atomic Energy Authority workforce'.

Whilst not a study of the general mortality it may give some information to respond with.

Regards
Ross 

-----Original Message-----
From: radsafe-bounces at radlab.nl [mailto:radsafe-bounces at radlab.nl] On Behalf Of Steven Dapra
Sent: 08 July 2008 21:18
To: radsafe at radlab.nl
Subject: Re: [ RadSafe ] Nuclear mortality

July 8

         I did a Google search for "Atomic veterans" "mortality study" and was referred to an online book by the National Academy Press.  The link below will take you to page one of the front matter.  This is the material on page one:
Mortality of Veteran Participants in the CROSSROADS Nuclear Test Medical Follow-up Agency INSTITUTE OF MEDICINE by J. Christopher Johnson Susan Thaul William F. Page Harriet Crawford with oversight from the Institute of Medicine Committee on the CROSSROADS Nuclear Test NATIONAL ACADEMY PRESS Washington, D.C.
1996

http://books.nap.edu/openbook.php?record_id=5428&page=R1

         I would recommend reading the Summary, the Study Rationale, and Other Studies of the Human Health Effects of Radiation Exposure.  Of course, this is only about the CROSSROADS test, however it should get you started.
The first Reference is to Beebe GW and Simon AH. Ascertainment of mortality in the U.S. veteran population. American Journal of Epidemiology 89:636---643, 1969; which will doubtless give a broader picture.

Steven Dapra

At 02:07 PM 7/8/08 -0400, Joel Cehn wrote:
>Regarding "Numerous medical studies in the last 50 years clearly 
>establish that when large populations are exposed to..."? is anyone 
>aware of a mortality study of "atomic veterans?"? I was asked about 
>that the other day.
>
>
>Joel I. Cehn, CHP
>Oakland, California

_______________________________________________
You are currently subscribed to the RadSafe mailing list

Before posting a message to RadSafe be sure to have read and understood the RadSafe rules. These can be found at:
http://radlab.nl/radsafe/radsaferules.html

For information on how to subscribe or unsubscribe and other settings visit:
http://radlab.nl/radsafe/ Internal Virus Database is out of date.
Checked by AVG - http://www.avg.com
Version: 8.0.138 / Virus Database: 270.4.5/1535 - Release Date: 04/07/2008
5:03 PM

_______________________________________________
You are currently subscribed to the RadSafe mailing list

Before posting a message to RadSafe be sure to have read and understood the RadSafe rules. These can be found at: http://radlab.nl/radsafe/radsaferules.html

For information on how to subscribe or unsubscribe and other settings visit: http://radlab.nl/radsafe/