Sample Size calculations with sub-groups

I am in a situation wher we need to recruit 1600 people with a particular condition (calculated so that we can detect a 5% change following intervention). We also want to test the intervention on a particular sub-type (drug-users), where we expect a higher prevalence of the condition, and want to still be able to detect the 5% change.

My questions are:
- In calculating the sample size, do I treat the sub-group as part of the larger group, or do I calculate the two sample sizes separately and then add them together for the total sample size. Presumably this would mean that we'd need another 1600 people for the sub-group, which would mean significant over-sampling.
- Do we conduct disproportionate stratified samplingf where the population is stratified according to drug/non-drug use. If this is the way to go, how is the sample size calculated?

How is this type of situation usually approached????