[ RadSafe ] Incidence vs death question, was: Exposed " -had lower incidences of all cancers - "

[Keith Welch]

Folks,
I am not an epidemiologist and have no experience in that field.  But 
recently, partly due to the posts here, I have been wondering about 
this.  Maybe I just haven't thought it through well enough.  It seems on 
its face that using cancer incidence rates would be preferable to 
mortality, due in part to the issue of changes over time in cure rates, 
but also because it would seem to help correct for the healthy worker 
effect (incidence rate is not as affected by the availability of health 
insurance or treatment as mortality rate) - and possibly the "rich 
victim effect", which I have not heard many people talk about, but 
assume must be confounding; the difference in cure rates in different 
socio-economic classes.  I would suppose that could probably be dealt 
with by careful cohort selection.  At any rate, I've heard that the 
shipyard worker study was flawed due to the following: (1) screening for 
nuclear workers at the shipyards disqualified people with family history 
of cancer, and (2) removal of people from nuclear worker status (and 
therefore, presumably from candidacy for the study?) in the event they 
were diagnosed with cancer during employment.  Are either of these based 
in fact?

Keith Welch
Jefferson Lab