Dear all,
have you ever been in a situation where subjects in a Case-Control or
Cohort study are not independent?
This is the typical case of population isolates where almost all the
subjects or a very high percentage of them are strongly related.
I have a population of about 30,000 inhabitant, clustered in villages of
400-5000 people each, with strong familial aggregation.
Which strategies do you suggest to study risk factors for a disease in a
population like this?
How do you think non-independence could affect the effect estimates (I
guess power is reduced, but to which extent?)? Would you suggest any way
to face the problem?
Thanks a lot!
Cristian Pattaro
Unit of Genetic Epidemiology
Institute of Genetic Medicine
__________________________________________
EURAC research
Viale Druso/Drususallee 1
39100 Bolzano/Bozen - Italy
Tel. +39 0471 055 527 - Fax +39 0471 055 599
Email: [log in to unmask] <mailto:[log in to unmask]>
Website: www.eurac.edu <http://www.eurac.edu/>
|