I have a database with 1500 variables, and 700446 observations...
For example: Sex, dissability, work, abuse, salary, etc.
I would like to analize just deafness (as a dissability) in women with all the other, to see the relation between that women deaf and work-abuse-etc...
That's one high-dimensional data set. If it's stored in a SQL database, I can give you a recipe to extract just the fields of interest. Unless you have a really quite large RAM, importing the entire dataset and narrowing it down in R is likely going to crash the program.
So to start @Rosana, can you tell me what format the source data is in, and we can work from there?