Sometimes you may only be interested in describing a population (e.g., you want a breakdown of your practice by sex, age, race, urban dweller versus rural dweller, etc.). You might even go into hierarchies of descriptive information (e.g., of all my male patients over the age of 35, how many are diabetic). This may even be focused on a particular problem (e.g., how many diabetic patients do I have, and what is their breakdown by age, race, sex, etc.). This is called a descriptive study.
But sometimes you can actually conduct what might be called an "artificial experiment". You are able to identify one group that has received (or will be receiving) a particular treatment and one group with a similar condition that is receiving an alternative treatment, but for whatever reason(s) you have not been able (or will not be able) to influence who receives which treatment. Nevertheless, you can still gain some knowledge from such a situation. There are primarily two different ways this might be done: