Epidemiology: Case Control selection criteria

[bwbddb]

For you epidemiology buffs:
A recent practice question puzzled me:

If a physician were to design a study to determine if accupuncture can be used to induce labor, would he choose a case control or cohort study design (assuming only those options)?

The 'correct' answer to the practice question was case control. However, my understanding of case control is that you always choose cases based on disease/condition (in this case, labor) and look back to examine exposure (accupuncture). With a rare exposure, this isn't effective. Wouldn't a cohort study be a better design, or can you choose cases based on exposure? The complete question can be found in 'Blueprints Clinical Cases: Obstetrics and Gynecology' 2006 pg 132 which is available on Google Books.