Does anyone have a mnemonic or something like that to remember the difference between the two and which is retro and which is prospective? Any help appreciated!
Case control: compares what two separate populations look like right now - I always think of it as a snapshot or a instant in time, and therefore you can't establish causation with case/control. Cheap, easy to conduct. Prospective cohort: you look at one group and watch them. People fall out on risk factors or prevalence of disease. Think about looking into the future and letting God sort them out. Goes towards establishing causality and is expensive, time-consuming to conduct, but the brass ring of studies. Retrospectives: kinda like prospective cohorts, but you're looking at old data. The others need subjects, retrospectives need patient charts.
Also remember that in a cohort study, no one has the disease of interest, you're looking at risk factors to see who might develop the disease.
RXNMAn's definition of Case Control looks very similar to Transversal Study which is at a particular point in time???
I might be wrong, but I certainly thought case-control was a retrospective study. You look at people with the disease, and see what risk factors they had. Cross-sectional is the snapshot in time you speak of. For analysis, remember CCOR (case-control: odds-ratio) and CRR (cohort: relative risk)... the rule is that you have one and only have one letter pairing, not two. CCRR doesn't work, nor does COR.
The easiest way is to think about the words: Case-Control: you have a group who is the CASE (with the dz) and a group who is CONTROL(w/o the dz) now you want to look at what the hell these two groups did in the past (risk factors) and see if the cases did anything different than the controls. Therefore these are retrospective studies - use Odds ratio Cohort: you gather a bunch of people who have been doing (or will do) the same behavior. They are cohorts - because they have the same behavior. Then you see which ones developed the dz - now you can determine whether that behavior was actually a risk factor or not. These can be retrospective, where all data has already been collected, everything is said and done, you just go through the charts. Or prospective, where you watch the cohort over a period of time.
I think you have the right idea about Case control, but it is a retrospective study. Case-control involves looking at a group with (case) and without (control) the disease and looking for risk factors in the past (retrospective). The two groups are determined based on the presence or absence of the disease at a certain point (as you had said), but the analysis (odds ratio) involves looking retrospectively for the presence or absence of exposure to a risk factor. -Hope this helps
A cross-sectional will also get you a snapshot in time. Case-control - take a bunch of people with a disease, take a bunch of people without the disease (in academia, this means the grad students down the hall), and apply some test to both groups. E.g. "Our new lung CA test had a sensitivity of 0.90 comparing between cases and controls." No causation, but correlation can be inferred from these results. Cross-sectional - take a bunch of people, give them all surveys, a test, whatever, and analyze the results. You can say things like, "The people who scored highest on lifetime use of ectasy had the lowest fMRI results for areas 1, 2, and 5." There's no comparisons between groups, there's just how a group looks. I will have to think about your mnemonic - I always get RR and OR mixed up. On cohorts - people are grouped by one common trait - birthday, a risk factor, activity, or being from Boston. It doesn't matter. Then you observe them for (usually) 5 years. You define your outcome measures (usually death or develpoment of the disease of interest) at the beginning. Think of the Framingham study. I can't think of a way that you could do a retrospective case-control study. You could look at all of the charts of people admitted to a hospital over a given time and how their diagnoses and outcomes compared to another group's, but you might as well just call it a retrospective cohort study at that point. EDIT: OK, I guess I got some things confused. Taking a look at this, I need to brush up on this subject myself.
Had 3 case-control/cohort questions on my exam 6/16: One was a little strange involving weight control in bus conductors and drivers The other two - very obvious...
