Re: Source of cancer data

[RuthWeiner@AOL.COM]

In a message dated 12/31/01 4:20:51 PM Mountain Standard Time, JohnWi@law.com writes:



> In a case control study, the control of confounding such as smking 
>
> rate and duration is performed at the level of the individual.  The 
>
> interviewer asks each respondent about thir smoking history.  I used 
>
> smoking as an example, but it also relates to other data that is 
>
> collected.  In the ecologic study, Cohen used sales tax to infer 
>
> smoking rates for the county


I do not believe asking individuals about their smoking history differentiates a case control study from other types of epidemiologic studies.  Many epidemiological studies are made by asking individuals questions.  I mean, one can hardly ask a questions of an entire group.  It is my understanding, though, that the "cases" and the "controls" are both groups, and the study considers the group parameters, not the individual ones.   However, I am not an epidemiologist and I would appreciate a critique of my comments from someone on RADSAFE with epidemiology credentials and experience.  


Some years ago, I helped with what I understood to be a case control study of children who lived in the "fallout" of the ASARCO smelter in Tacoma, WA.  The group of Tacoma children was matched with a group of children who lived in a similar industrial area in Seattle, had the same spread of ages and grades in school, and were exposed to essentially the same air pollutants except for the airborne arsenic that the Tacoma children were exposed to.  The groups were approximately the same size.  The careful control of confounding factors (ages, group size, numbers of boys and girls, environmental exposure) are what (I thought) made this a case control study.  The focus of the study was health differences between the Tacoma and Seattle groups, to see if any health differences could be attributed to exposure to airborne arsenic.


Incidentally, responses to questions about individual smoking, dietary, and alcohol use histories are notoriously unreliable.  For example,  my husband and I occasionally have wine with dinner, but I can't tell you to better than +/- 30% how many glasses of wine I have had in the past month.  One would get a more reliable number by checking the wine entries on my supermarket shopping register tapes.


Ruth Weiner, Ph. D.

ruthweiner@aol.com