biostats concepts plz help

[sadaca]

---

Members don't see this ad.

hey guys, For incidence you have to subtract the ppl that have the disease, but do u also subtract the ppl that died from that dx??
I know its new cases/ total popl at risk (subtract ppl with dx). I had a uworld q on this and I am not sure what do about the dead ppl if they are included in here??

also does anyone know how to differentiate between Effect modification and confounding ? I am attaching the definition of Effect modification. If anyone knows any examples for confounding bias that would help me alot.

Thanks alot

 

[Nephronlearner]

I am not sure I know what you mean, especially without seeing the question, but 'incidence' is exactly that - incidence. The outcome should not matter.
So all that matters when it comes to incidence is what number of people got the disease.

So if you have 1 million people and 20 cases the incidence is 20/1,000,000 = 2/100,000.
However, it appears that there are 2 variables - smoking and the drug in the question.
Again, I am not sure what exactly the question is asking but I assume you just need to add in the 2nd variable. (smoking/drug)

 

[sadaca]

Hey thanks for your help

 

[Missorleans]

when you stratify by an effect modifier, the effect sizes are different. When you adjust for a confounder, you'd find there is no change in effect size.
So in the example you posted, when you adjust the effect measurement for smoking, there would be no difference between the treatment groups.

 

[sadaca]

hey missorleans! Thanks alot, question for you-you said that for effect modifier the effect sizes are different but then later you said for my example(on effect mod) there would be no difference? I am confused by that. I understand that Effect modifier is not a bias so would there a difference or not? Do u have any examples for confouding? thanks alot

 

[Missorleans]

hey missorleans! Thanks alot, question for you-you said that for effect modifier the effect sizes are different but then later you said for my example(on effect mod) there would be no difference? I am confused by that. I understand that Effect modifier is not a bias so would there a difference or not? Do u have any examples for confouding? thanks alot

I think for the example there was a difference in effect size when the stratified according to smoking status? So that would make your example one of effect modification. If there was no difference, it would be confounding.
An example of confounding would be finding an association between coffee drinking and stomach cancer. But those who drink coffee might also be more likely to smoke, so when you stratify the data by smoking status, the effect estimates would actually be the same. I.e. There isn't really an association between coffee drinking and stomach cancer, it just looks that way because you did not control for smoking status.

 

[sadaca]

hey thanks alot. yep that makes sense now! Thank you!