"real" epidemiology?

[Michael Stabin <stabin@npd.ufpe.br>]

> The study, conducted over a two-year period in two separate regions
> of Sweden, evaluated a total of 233 patient cases with verified brain
> tumors. Each of these patient cases was matched to two controls, or
> healthy subjects (466 in total), based on similar sex, age, and
> geography. Eight of these 233 patients had recurrent brain tumors and
> were excluded from the study together with their matched controls.
> The cases and controls were evaluated for exposure to a variety of
> possible cancer risks through questionnaires and additional telephone
> interviews.

Could I elicit some comments from those who really know epidemiology on the
list as to the real value of this kind of research in forming scientific
opinion?  I have long felt that we are going through a period of time in
which we can do wonderful things and terrible things with computers.  This
kind of questionnaire-based epidemiology, which shows no underlying
biological mechanism, but simply points out if certain variables in
population groups are correlated - isn't that why we have studies that one
week say that this or that (caffeine, high fiber diets, radiation from cell
phones) raise/reduce cancer rates and the next week we hear the contrary?
Is this good science?  My understanding of epidemiology is relatively
superficial, so I'm asking, not opining (for a change!).


Michael Stabin, PhD, CHP
Departamento de Energia Nuclear/UFPE
Av. Prof. Luiz Freire, 1000 - Cidade Universitaria
CEP 50740 - 540
Recife - PE
Brazil
Phone 55-81-271-8251 or 8252 or 8253
Fax  55-81-271-8250
E-mail stabin@npd.ufpe.br

"Quantum Mechanics: The dreams stuff is made of"
- Steven Wright


************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html