Observational Studies (Study Designs): Summary

00:01 So, when are you going to use which designs? This is kind of fun. So remember where I mentioned that case-control studies are useful when my exposure status is common and the outcome is rare and I mentioned that the cohort designs are useful when the exposures are rare and the outcomes are common, let's put that information to use. Let's say we're doing a short-term study of current uncommon exposures and common outcomes. Obviously that's going to be a prospective cohort, because the outcome is common, I know I'm going to manifest some cases. If I'm doing a short or long-term study of historic uncommon exposures and common outcomes, it's still a cohort design, but it's happening in the past, so it's a retrospective. If I'm doing an outbreak investigation, well this is classic, almost all outbreak investigations are case-control scenarios. Because by the time you arrive, the outbreak has already happened, you have some people who have the disease and you find some people who don't have the disease and you look back to see what caused the disease likely. Think about a picnic, let's say some people going to a picnic and some of them get sick, some kind of gastrointestinal disease.

01:10 The outbreak investigator, let's say it's you, has to arrive and figure out which food in the picnic made them sick. While you arrive, you find some people who have the illness and some people who don't have the illness and you ask them, what foods did you eat in the picnic and you compare those results and you figure out which food was likely the exposure that caused the outbreak. That's a classic case-control scenario. And instantaneous surveys, well, that's a cross-sectional design obviously.

01:37 So let's work through some specific examples. Is there an association between climbing Mount Everest and getting diabetes? Okay, is climbing is Mount Everest rare or common? I think it's kind of rare, I don't know anybody who has ever climbed Mount Everest, maybe you have, and gained diabetes, unfortunately that is quite common. So the combination of an uncommon exposure and a common outcome is a classic case for a cohort design. Is there an association between left-handedness and getting mad cow disease? Well mad cow disease in human is quite rare; left-handedness is quite common, so that's a case for a case-control design, again an uncommon outcome. Is there an association between left-handedness and gender? Well the important considerations here are that left-handers and gender, male or female, are very common, they're also tend to be unchanging, so it doesn't matter which I ascertained first or afterwards, that's the case for a cross-sectional study, because the timeframe is irrelevant.

02:39 What factors were likely responsible for the salmonella outbreak at the office Christmas party? That's an outbreak investigation. As we established, case-control designs are classically appropriate for outbreak investigations. And was there an association between working on the nuclear bomb project in World War II and developing cancer five years later? Well that took place in the past, so we know it's going to be a retrospective or historic investigation and we know that cancer is not that uncommon, but working on a nuclear bomb project was uncommon. So that's going to be a case for a cohort, in other words, a retrospective cohort. So when we want to minimize the influence of extraneous factors that may confound our