specifying independence of random effects?

Hello

I'm trying to specify a mixed-effect model describing "unmet NEED" (of  
families with disabled children) as a function of IMPAIRMENT (i.e.  
intrinsic severity of the child's condition) and AREA of residence  
(factor with ~15 levels). The impairment measure is a unidimensional  
variable but it needs "calibrating" for separate types of disability  
(autism, cerebral palsy etc) represented by another factor, DISCAT

In lmer syntax I have

NEED ~ IMPAIRMENT | DISCAT + 1 | AREA

but I want to specify independence of random effects. I'm aware of the  
pdDiag options in the nlme library but unsure how to specify the model  
in lme syntax?

Thank you

Rob