Statistics: Cohort or case control

[Augurey]

The case is as followed (and sorry for the bad translation):

"There is much that indicate that exposure to different chemicals can increase the risk for renal cancer. Would you use a cohort study or a case control study to answer the question?"

There is only one best answer. I hope that some of you can explain to me how this is so obvious. For me it's not.

---

Members don't see this ad.

 

[OrthoTraumaMD]

A cohort study follows patients with exposure to see if they develop a consequence (disease).

A case control study follows patients with and without the disease and looks to see what the exposure was in each.

Cohort studies are difficult to conduct but better to prove causation, because case control studies are prone to bias. So I would say since you are looking for the answer to "will the exposure cause the cancer," the study you want is a cohort.

 

[Augurey]

A cohort study follows patients with exposure to see if they develop a consequence (disease).

A case control study follows patients with and without the disease and looks to see what the exposure was in each.

Cohort studies are difficult to conduct but better to prove causation, because case control studies are prone to bias. So I would say since you are looking for the answer to "will the exposure cause the cancer," the study you want is a cohort.

Thank you for your reply. That was the same as I was thinking, but the correct answer is supposed to be case control. The explanation is that cohort is time consuming and you don't know what to set as the exposures. I was a little confused when I read this.

 

[xffan624]

Thank you for your reply. That was the same as I was thinking, but the correct answer is supposed to be case control. The explanation is that cohort is time consuming and you don't know what to set as the exposures. I was a little confused when I read this.

Think of that question as an outbreak scenario. A cohort is good when you have an idea of the exposure that you're looking for. Case control is effective when you really have no idea of the exposure (different chemicals is a pretty broad statement). You're basically fishing. You have the outcome, but not how you got there. So you match folks that do and don't have the outcome and then ask about a bunch of different exposures. The question is really not specific enough, but I guess that's the point they are trying to make. I'm not sure I would have gotten it right either, but I understand what they're saying.

 

[Augurey]

Think of that question as an outbreak scenario. A cohort is good when you have an idea of the exposure that you're looking for. Case control is effective when you really have no idea of the exposure (different chemicals is a pretty broad statement). You're basically fishing. You have the outcome, but not how you got there. So you match folks that do and don't have the outcome and then ask about a bunch of different exposures. The question is really not specific enough, but I guess that's the point they are trying to make. I'm not sure I would have gotten it right either, but I understand what they're saying.

Ok, thanks. I understand that point of view as well. At the same time, when I read the question I thought that "different chemicals" wasn't that broad. I thought that they had a group of chemicals (that were known), and that these chemicals would be the exposure. Is that an understandable assumption or completely wrong?

I basically want to know if the correct answer is up to debate based on the information given in the question.

 

[xffan624]

Ok, thanks. I understand that point of view as well. At the same time, when I read the question I thought that "different chemicals" wasn't that broad. I thought that they had a group of chemicals (that were known), and that these chemicals would be the exposure. Is that an understandable assumption or completely wrong?

I basically want to know if the correct answer is up to debate based on the information given in the question.

Think of it like a church picnic, the classic case control scenario. Different foods were available at the picnic. Not everybody ate all the foods. You use a case control to compare which ones the sick vs. non-sick ate. Similar comparison. This is assuming you are going back in time. That should be better to clarified in this question. Case control studies are always backwards. Cohorts are either backwards or forwards, but if it's a backwards cohort you probably need a specific exposure you are looking for.

 

[Augurey]

Think of it like a church picnic, the classic case control scenario. Different foods were available at the picnic. Not everybody ate all the foods. You use a case control to compare which ones the sick vs. non-sick ate. Similar comparison. This is assuming you are going back in time. That should be better to clarified in this question. Case control studies are always backwards. Cohorts are either backwards or forwards, but if it's a backwards cohort you probably need a specific exposure you are looking for.

Thank you for the useful example.

I should have mentioned that it's a prospective cohort.

 

[xffan624]

Thank you for the useful example.

I should have mentioned that it's a prospective cohort.

??? You just said the answer was case control. How can it be both? Case control studies are always retrospective.

 

[Augurey]

??? You just said the answer was case control. How can it be both? Case control studies are always retrospective.

No, I meant that the choice was between case control and prospective cohort.