Private Wang Fire wrote:Unome wrote:Private Wang Fire wrote:
Cohort presumes that you are tracking the entire exposed/unexposed population to the outcome, whether they develop the disease or not. Case-control starts with the outcome, so when you go back towards whether they were exposed/unexposed, you don't have the whole population, just a sample. Does this make sense? I'm not a very clear explainer
So why can a cohort study generalize a determined relative risk to a broader population of people? (or can it just not?)
The "population" in the case of a cohort is the entire exposure group, as in everyone who was exposed to that specific exposure at the specific time and place in the case study, which means relative risk can be used.
In the case control, the cases or controls you select initially represent only part of the total "population" that you're studying - population in this case would be defined to strictly those with nearly identical (in an ideal case) characteristics and the only thing that sets the two groups in the study apart would be whether they were exposed or not.
So not like a broader population of people, just like the population that is involved in the disease 'event' or whatever.
Another thing you might want to consider when thinking about this is that cohort studies are trying to find a disease, and case-control is trying to find a exposure