If there is such an association, how would a study differentiate between the underlying condition causing the increased risk, and the medication itself?

Edit: smokers have reduced rates of Parkinson's disease, and nicotine is a stimulant.